Dermatological imaging system guided by 3D mapping

Abstract

An imaging system including an image acquisition assembly operable to capture photographic images of a subject surface; an actuator coupled to the image acquisition assembly; a location sensor operable to detect location and orientation of the image acquisition assembly while capturing overlapping images; a three-dimensional (3D) mapping assembly that obtains a 3D topography map representative of a 3D surface geometry of the subject; an illumination assembly operable to illuminate a camera field of view so that the captured photographic images include medical dermatoscopic information; and a computing unit coupled to the image acquisition assembly, the actuator, and the 3D mapping assembly. The computing unit is operable to: develop a scanning plan based on the 3D topography map; execute the scanning plan on-the-fly via the actuator and trigger the image acquisition assembly; and deblur at least one sequence of significantly overlapping images utilizing the 3D topography map.

Description

DERMATOLOGICAL IMAGING SYSTEM GUIDED BY 3D MAPPINGCROSS-REFERENCE TO RELATED APPLICATIONSThis application claims the benefit of U.S. Provisional Application No. 63 / 644,629, filed May 9, 2024; U.S. Provisional Application No. 63 / 737,906, filed December 23, 2024; and U.S. Provisional Application No. 63 / 783,613, filed April 4, 2025, the disclosures of which are incorporated by reference herein in their entireties.SUMMARY

[0001] The present disclosure is directed to methods, systems, and apparatuses for dermatological imaging that are guided by three-dimensional (3D) mapping.

[0002] In one embodiment the system for dermatologic imaging performs a fast full scan of at least a portion of a mammal subject body surface, guided by a 3D scanning plan and updated by a location sensor.

[0003] In another embodiment, the dermatologic imaging is guided by a 3D scanning plan, and the captured images are deblurred by suitable software.

[0004] In another embodiment, the system for dermatologic imaging captures images on- the-fly.

[0005] In some aspects, the techniques described herein relate to an imaging system including: an image acquisition assembly including a camera and imaging optics configured to capture photographic images of a skin surface of a mammal subject disposed on a support structure while executing a scanning plan; an actuator coupled to the image acquisition assembly, the actuator configured to move the image acquisition assembly over the subject at a near-nominal working distance; a location sensor configured to detect location and orientation of the image acquisition assembly while capturing overlapping photographic images, wherein the location sensor includes at least one light source configured to direct an electromagnetic beam to a target surface; a three- dimensional (3D) mapping assembly that obtains a 3D topography map representative of a 3D surface geometry of the subject; an illumination assembly configured to illuminate a camera field of view so that the captured photographic images include medicaldermatoscopic information; and a computing unit coupled to the image acquisition assembly, the actuator, and the 3D mapping assembly, the computing unit configured to: develop a scanning plan based on the topography map, the scanning plan including a traversal path of the image acquisition assembly and a camera triggering order along the traversal path, the traversal path maintaining the near-nominal working distance between the skin surface and the image acquisition assembly and a view axis of the image acquisition assembly is near-normal to the viewed captured skin surface; execute the scanning plan on-the-fly via the actuator and trigger the image acquisition assembly to obtain at least one sequence of significantly overlapping images along the traversal path; and deblur at least one sequence of significantly overlapping images utilizing the 3D topography map.

[0006] In some aspects, the techniques described herein relate to a method including: positioning a mammal subject on a support structure of an imaging system so that a skin surface of the mammal subject is visible by an image acquisition assembly of the imaging system including a camera configured to capture photographic images of the mammal subject on-the-fly; obtaining a topography map representative of a three-dimensional (3D) surface geometry of the mammal subject using a 3D mapping assembly of the imaging system; developing a scanning plan based on the topography map, the scanning plan including a traversal path of the image acquisition assembly and an image capture event triggering order along the traversal path, the traversal path maintaining a near- nominal working distance between the skin surface and the camera and further maintaining optical axis orientation of the camera near normal to skin FOV of the surface of the mammal subject during each image capture; executing the scanning plan via an actuator, a location sensor, and the image acquisition assembly; simultaneously illuminating the subject and capturing significantly overlapping photographic images of the subject at each image capture event while executing the scanning plan, wherein the captured photographic images include captured dermatoscopy images and captured skin images; converting a location sensor readout from the location sensor into image acquisition assembly five degree of freedom (5DOF) coordinates for each captured photographic image; deblurring each captured image utilizing a respective section of the topography map; and stitching the deblurred captured photographic images into acontinuous high-resolution dermatoscopy images and continuous high -resolution skin images.

[0007] In some aspects, the techniques described herein relate to an imaging system including: an image acquisition assembly including a camera and imaging optics operable to acquire photographic images of a mammal (e.g., human) subject while executing a scan plan; an actuator coupled to the image acquisition assembly, the actuator operable to move the camera over the subject at a near-predetermined, nominal working distance; a location sensor operable to detect location and orientation of the camera while capturing overlapping images; a three-dimensional (3D) mapping assembly that obtains a topography map representative of a 3D surface geometry of the subject; an illumination system operable to illuminate a camera field of view so that the captured photographic images include medical dermatoscopic information; and a computing unit coupled to the image acquisition assembly, the actuator, and the 3D mapping assembly, the computing unit operable to: develop a scanning plan based on the topography map, the scanning plan including a traversal path of the image acquisition assembly and a camera triggering order along the traversal path, the traversal path maintaining near-nominal working distance between the subject and the imaging system so that an axis of the imaging system is near-normal to the captured subject surface; execute the scanning plan via the actuator and triggering the image acquisition assembly to obtain overlapping images along the traversal path; and combine the overlapping images taken by the image acquisition assembly into one or more high-resolution dermatoscopy images.

[0008] In some aspects, the techniques described herein relate to a method, including: positioning a subject on a support structure so that a surface of the subject is visible by an image acquisition assembly including a camera operable to acquire photographic images of the subject; obtaining a topography map representative of a three-dimensional (3D) surface geometry of the subject using a 3D mapping assembly; developing a scanning plan based on the topography map, the scanning plan including a traversal path of the image acquisition assembly and a camera triggering order along the traversal path, the traversal path maintaining a nominal working distance between the subject and the camera and further maintaining optical axis orientation near normal to skin field of view during each image capture; executing the scanning plan via an actuator and the imageacquisition assembly to obtain overlapping images along the traversal path; and combining the overlapping images taken by the image acquisition assembly into a sequence of high-resolution dermatoscopy and skin images.

[0009] In some aspects, the techniques described herein relate to an imaging system including: an image acquisition assembly including a camera and imaging optics operable to capture photographic images of a subject surface of a subject while executing a scanning plan; an actuator coupled to the image acquisition assembly, the actuator operable to move the camera over the subject at a near-nominal working distance; a location sensor operable to detect location and orientation of the image acquisition assembly while capturing overlapping images; a three-dimensional (3D) mapping assembly that obtains a topography map representative of a 3D surface geometry of the subject; an illumination assembly operable to illuminate a camera field of view so that the captured photographic images include medical dermatoscopic information; and a computing unit coupled to the image acquisition assembly, the actuator, and the 3D mapping assembly, the computing unit operable to: develop a scanning plan based on the topography map, the scanning plan including a traversal path of the image acquisition assembly and a camera triggering order along the traversal path, the traversal path maintaining the near-nominal working distance between the subject and the image acquisition system so that an axis of the image acquisition assembly is near-normal to the captured subject surface; execute the scanning plan on-the-fly via the actuator and trigger the image acquisition assembly to obtain at least one sequence of significantly overlapping images along the traversal path; and deblur at least one sequence of significantly overlapping images utilizing the 3D topography map.

[0010] In some aspects, the techniques described herein relate to a method including: positioning a subject on a support structure of an imaging system so that a surface of the subject is visible by an image acquisition assembly of the imaging system including a camera operable to acquire photographic images of the subject; obtaining a topography map representative of a three-dimensional (3D) surface geometry of the subject using a 3D mapping assembly of the imaging system; developing a scanning plan based on the topography map, the scanning plan including a traversal path of the image acquisition assembly and a camera triggering order along the traversal path, the traversal pathmaintaining a nominal working distance between the subject and the camera and further maintaining optical axis orientation near normal to skin FOV during each image capture; executing the scanning plan via an actuator of the imaging system, the location sensor, and the image acquisition assembly to obtain significantly overlapping images along the traversal path; deblurring each captured image utilizing the respective section of the topography map; and combining the overlapping images taken by the image acquisition assembly into a sequence of high-resolution dermatoscopy and skin images.BRIEF DESCRIPTION OF THE DRAWINGS

[0011] The discussion below references the following figures, wherein the same reference number may be used to identify the similar / same component in multiple figures.

[0012] FIG. l is a block diagram of an epidermal imaging system according to example embodiments;

[0013] FIGS. 2A and 2B are block diagrams of an image acquisition assembly according to an example embodiment, where FIG. 2A illustrates a light source and camera of the assembly, and FIG. 2B illustrates a polarizer and image sensors of the assembly;

[0014] FIG. 3 is a flowchart of a process according to an example embodiment;

[0015] FIGS. 4A and 4B are diagrams showing how three-dimensional subject body surface data is captured according to an example embodiment, where FIG. 4A illustrates the epidermal imaging system of FIG. 1 with a location sensor that includes a LiDAR device, and FIG. 4B illustrates the LiDAR device;

[0016] FIGS. 4C and 4D are diagrams showing how three-dimensional subject body surface data is captured according to an example embodiment, where FIG. 4C illustrates the epidermal imaging system of FIG. 1 with a location sensor that includes a projector and a screen, and FIG. 4D illustrates the projector;

[0017] Fig. 4E is a diagram illustrating intermediate FOV 3D depth maps, generated in the process of approaching a final depth map for a specific captured image;

[0018] FIG. 4F is a schematic plan view of a target surface of the system of FIG. 4A;

[0019] FIG. 5 is a flowchart illustrating an image deblur process that also stitches the deblurred images into a continuous image;

[0020] FIGS. 6A and 6B are diagrams showing aspects of a scanning plan according to example embodiments, where FIG. 6A is a schematic diagram of an imaging scanning plan, and FIG. 6B is a schematic diagram of the image acquisition assembly of FIG. 1 executing the imaging scanning plan;

[0021] FIG. 7 is a diagram showing signals generated by a scanning plan according to an example embodiment;

[0022] FIG. 8 is a block diagram of a computational unit according to an example embodiment;

[0023] FIG. 9 is a flowchart of a process according to an example embodiment.

[0024] FIGS. 10A and 10B are diagrams showing how three-dimensional subject body surface data is captured according to another example embodiment, where FIG. 4A illustrates the epidermal imaging system of FIG. 1 with a location sensor, and FIG. 4B illustrates the location sensor; and

[0025] FIG. 11 is a flowchart of another process according to an example embodiment.DETAILED DESCRIPTION

[0026] The present disclosure is generally related to dermatological imaging, such as total body photography used to monitor skin health and / or detect conditions such as skin lesions, cancers, etc., e.g., dermatoscopy imaging. As used herein, the terms “dermoscopy” or “dermatoscopy” refer to the examination of the skin using skin surface microscopy, and is also called “epiluminoscopy” and “epiluminescent microscopy.” Total body photography involves capturing high-resolution photographs of the entire skin surface. Combined with other advances in dermatological imaging, these procedures can significantly improve the early detection, diagnosis, and monitoring of various skin conditions.

[0027] Unlike other forms of medical imaging (e.g., X-rays, computed tomography scanning), a body surface photography scan obtains high resolution images at close range of a surface (e.g., the epidermal surface of a human body) that has complex geometry. Therefore, the maneuvering and focal adjustment of a high-resolution camera along this geometry can be time consuming. For example, in some experimental dermatologicalimaging devices, the working distance is 30-60 cm. The image acquisition camera is aligned with a specific region of interest and takes at least one high resolution image of that region. This alignment and acquisition are repeated over some or all of the body.

[0028] An imaging system that significantly reduces scanning and image acquisition times can decrease labor costs, e.g., the time spent by practitioners in monitoring the process. Reducing acquisition time can also reduce subject discomfort caused by having to remain motionless for a long period of time. Both of these factors can increase the adoption of total body photography and provide its benefits of early detection of skin disease to a larger population.

[0029] In various embodiments, an image acquisition assembly is described herein that is capable of fast, on-the-fly image acquisition of epidermal images. In one or more embodiments, the assembly is configured to capture photographic images of the subject surface on-the-fly at a high image capture events rate of at least about 15 Hz. In one or more embodiments, the assembly is configured to capture photographic images of the subject surface on-the-fly at a high image capture events rate of no greater than about 600 Hz. The proposed device quickly provides a scan of a mammal subject’s body skin surface in, e.g., 3 minutes or fewer for a full body scan. These images can be obtained in a standardized way (e.g., controlled light and acquisition distance) such that the images can be readily post-processed to increase quality of the images and provide more reliable detection of known anomalies. In one or more embodiments, the system can provide skin images with high resolution (e.g., 5-10 microns), which can cover the whole body of the subject.

[0030] Center coordinates (XYZ) of each captured image pair can be provided. These coordinates can be utilized to calculate image acquisition assembly 5 degree of freedom (5DOF) data or coordinates. Such coordinates can be calculated for each captured image pair utilizing a location sensor. The sensor can utilize a system frame as a reference. The location sensor can mitigate encoder technology limitations and in turn simplify positioning of the image acquisition assembly.

[0031] The image acquisition assembly can simultaneously capture a skin image and a dermatoscopy image having similar fields of view. The images can be detail-rich and thus enable the construction of an improved depth map for each image. The depth map canenable proper accurate focus deblur of captured skin and dermatoscopy images and eventually enable the generation of high resolution skin and dermatoscopy images.

[0032] In one or more embodiments, the system can capture images at short range that can in turn enable the following: a) coaxial (repeatable) illumination; b) short illumination pulses, which resolve motion blur issues; and c) structured / spectral lighting that enhance skin structures and more importantly, sub-skin features.

[0033] Further, the use of depth maps can improve skin structures classification.

[0034] Fast image capture can enable significant overlap of images, which in general can enable enhanced skin / dermatoscopy image resolution. Further, the use of structured light modulation during adjacent image capture can also be possible.

[0035] Further, use of a Convolutional Neural Network (CNN) algorithm can be utilized to deblur the captured images. Use of these overlapping captured images can assist with CNN algorithm convergence.

[0036] In one or more embodiments, a metalens can be utilized with the image acquisition assembly of the system. The metalens may at least one of (a) reduce image chromatic aberration; (b) improve imaging resolution throughout the FOV; (c) assist with providing a depth map directly from chromatic dispersion of the metalens; (d) lower image acquisition assembly weight and inertia; or (e) enable compact structure that can also reduce inertia of the image acquisition assembly.

[0037] FIG. l is a schematic diagram that illustrates a medical skin or epidermal imaging system 100 according to one or more embodiments. In this particular embodiment, the system 100 includes a support structure 102 designed to accommodate a mammal subject 103 to be examined. The mammal subject 103 can be any suitable mammal, e.g., human, primate, etc. The support structure 102 may be a bed, ramp, chair or the like. Generally, the support structure 102 allows the subject 103 to comfortably stay still in a desired position, such as prone, tilted, upright, etc., for a reasonable amount of time, e.g., 5-10 minutes. A removable cover 101 may be used as a privacy cover, and can also be used to minimize external and stray light from entering into the skin surface or the imaging optics during the examination.

[0038] A positioning assembly 104 provides positioning and orientation of an image acquisition assembly 106. In one or more embodiments, the positioning assembly 104 canbe configured to substantially maintain a nominal working distance between a subject’s skin surface 105 and a front end of the image acquisition assembly 106. Further, in one or more embodiments, the positioning assembly 104 can be configured to assist the image acquisition assembly 106 in acquiring near-focused images of at least a portion of the viewed subject’s skin surface 105 during a scan. In this example, the positioning assembly 104 includes a first linear actuator 108 that moves along a first double rail 109 in the Y-direction. The first linear actuator 108 carries a second linear actuator 110 along a second curved rail 111. The second linear actuator 110 moves along the second curved rail 111, generally within the XZ-plane. The second linear actuator 110 carries a support 112 that holds the image acquisition assembly 106. The linear actuators 108, 110 may have motive elements such as motors and pinion gears that move together with the actuators along the rails 109, 111. In other embodiments, the actuators 108, 110 may be driven by belts, cables, or the like that are driven by fixed motors. The positioning assembly 104 also provides positioning of a 3D mapping assembly 114. The at least one of the linear actuators 108, 110 is configured to manipulate the image acquisition assembly 106 within five degrees of freedom.

[0039] The support 112 moves the image acquisition assembly 106 in the Z-direction in this orientation, or more generally in a radial direction relative to a center of the rail 111. The support 112 may also allow rotation about two axes, e.g., the X and Y axes. The positioning assembly 104 is coupled to a location sensor 115 that senses location and orientation of the image acquisition assembly 106, e.g., relative to a reference plane 120, which is fixed about the support structure 102. The location sensor 115 is used for closed- loop control of the positioning assembly 104 during the scan. In other embodiments, the positioning assembly 104 can use an open loop controller to control movement of the positioning assembly, e.g., using a map of input signals to locations in the reference frame. Such open loop mapping can be regularly updated, e.g., by moving to one or more reference locations and updating the mapping based on input readings (e.g., distance sensor) at the reference locations. The support 112 maintains the image acquisition assembly 106 such that its optical axis is near perpendicular to the current imaged surface, captured within the image acquisition assembly field of view (FOV). Other positioning apparatuses may be used instead of the illustrated linear-actuator-basedpositioning assembly 104. For example, a robot arm may be able to provide five degree- of-freedom (5DOF) positioning of the image acquisition assembly 106.

[0040] The 3D mapping assembly 114 collects three-dimensional (3D) geometry of at least a portion of the body side of the subject 103. The 3D mapping assembly 114 includes a stereo camera 116 that is configured to map the subject’s skin surface 105 in respect to a reference plane 120. An example of the stereo camera 116 is described, e.g., in “3D-imaging: A scanning light pattern projector” by Stokholm et. al., Applied Optics, vol. 55 no. 32 , pp. 9074-9083, 2016. A line array projector was coupled with a liquid lens for scanning a line array on a complex 3D object. Reconstruction of the scanned object was demonstrated.

[0041] The location and orientation of the stereo camera 116 relative to the reference plane 120 during each 3D mapping acquisition event can be used to locate each measured 3D map section within the reference plane, with its illustrated XYZ-coordinate system and its acquisition time. A computing unit 122 of medical skin imaging system 100 stitches the discrete acquired 3D maps sections into a continuous frozen body 3D map of at least a portion of the subject’s skin surface 105 facing the stereo camera 116. As used herein, the term “frozen body 3D surface map” means stitched continuous body 3D surface map. In certain aspects, the frozen body 3D surface map is stored by the computing unit 122 as a 2D array of points, whose XYZ coordinates are relative to the reference plane 120.

[0042] The 3D mapping assembly 114 may use other types of stereoscopic imagers in which one or more camera sensors capture images at different locations. Various techniques are known to perform this type of mapping, such as structure from motion (SFM). In some cases, the 3D mapping assembly 114 may include an independent positioning assembly to move the stereo camera 116 while performing the mapping. In other embodiments, the 3D mapping assembly 114 may use the same positioning assembly 104 as the image acquisition assembly 106. In other embodiments, the 3D mapping assembly 114 may be fixed relative to the support structure 102. In such a case, 3D mapping over an area may be performed by a scanning element (e.g., mirrors or solid- state steering in scanning light detection and ranging (LIDAR)) instead of the stereo camera 116.

[0043] The electronic components shown in FIG. 1 are coupled to the computing unit 122 configured to provide computing functions described elsewhere herein. The computing functions may include at least one of the following: (a) generating suitable scan plans for the stereo camera 116 and image acquisition assembly 106; (b) gathering measurement data or images from the location sensor 115 and calculating position and orientation of the image acquisition assembly 106; (c) gathering measurement data and images from the stereo camera and image acquisition assembly 106 and processing the data; or (d) sending focus adjustment data to the image acquisition assembly 106 lens. The computing unit 122 may include multiple processors, e.g., embedded controllers for positioning, digital signal processors for image acquisition and captured image deblur, central processing units for system control, subject interface, and the like. More specific details of the functions and arrangement of the computing unit 122 are described elsewhere herein.

[0044] In FIGS. 2A and 2B, diagrams show elements of the image acquisition assembly 106 according to an example embodiment. The image acquisition assembly 106 can use a coaxial illumination assembly 200 that is fixed to the support 112 of the image acquisition assembly in this example. The illumination assembly 200 includes at least one high-power light-emitting diode (LED) 204, a light confinement chamber 206, LED cooling means 208 (e.g., a fan), and a polarizing filter 210. The polarizing filter 210 (also referred to as a polarizer) may be switchable, and two or more different polarizers may be selected. Generally, the illumination assembly 200 provides a near-constant illumination field (e.g., pattern, spectrum, angular distribution) on the imaged skin surface 105 during the imaging part of the scan. The illumination assembly 200 may be designed for low weight, e g., to minimize loads on the actuators 108, 110 and support 112.

[0045] The image acquisition assembly 106 includes at least a camera 212 for capturing images of the skin surface 105 of the subject 103. The camera 212 includes at least one digital image sensor 214, such as a charge coupled device (CCD) or complementary metal-oxide semiconductor (CMOS) sensor. The image acquisition assembly 106, including a high-resolution telescope 216, is configured for capturing high resolution images of a small skin section within its FOV (“skin FOV”), where a skin FOV 127 (FIG. 1) has a specific geometry and orientation (e.g.. flat, perpendicular to the optical axisvector) , hereinafter, “aligned skin FOV,” and the skin FOV center is located at the working distance from the telescope 216. The imaging optics of the image acquisition assembly 106 can utilize a focus adjustment unit 218 configured for adjusting the focus of at least a portion of the aligned skin FOV. The camera 212 is shown fitted with a polarizer 220 having a polarization plane that may be crossed with a polarization plane of the lighting polarizer 210. In one or more embodiments, the polarizer 220 may have a 90 degree offset relative to the lighting polarizer 210 of the illumination assembly 200. The image acquisition assembly 106 may be designed for low weight, e.g., to minimize loads on the actuators 108, 110 and support 112.

[0046] The difference between the actual and planned distance measured between the FOV center on the subject’s skin surface 105 and the telescope 216 distal end according to a scanning plan 412 (FIG. 6A) may be corrected by the support 112 and complemented by the focus adjustment unit 218. The focus adjustment unit 218 may include any suitable optical elements or components, e.g., a controlled liquid lens 219.

[0047] The captured image may be blurred by one or more of the following effects: (a) the subject’s body 3D geometry, which can be exacerbated by the short working distance (typically 10 cm) of the image acquisition assembly 106; (b) image assembly 5DOF drift (i.e., distance, angular and lateral) from the scanning plan 412 during a scan; c) scan induced motion blur; or (d) natural body motion (e.g., body motion caused by a breathing cycle).

[0048] In one or more embodiments, the drift from the scanning plan 412 (including the telescope 216-skin FOV center distance) can be corrected by using the focus adjustment unit 218. In other embodiments, the computing unit 122 can provide the actual image acquisition assembly 106 5DOF coordinates relative to the reference plane 120 from location sensor 115 data. The computing unit 122 may use the actual 5DOF coordinates for sending a suitable focus correction signal to the focus adjustment unit 218.

[0049] The image acquisition assembly 106 may provide two types of images by dividing the light collected by the telescope 216 into at least two imaging paths that are directed towards two digital image sensors. The first imaging path includes the polarizer 220 set at cross polarization (relative to polarizer 210) and the image sensor 214. The second imaging path includes a visible light image sensor 222 and optional techniques thatdegrade the lighting polarization. Thus, image sensor 222 may provide high-definition skin surface information with minimal under-skin information.

[0050] In one or more embodiments, the image acquisition assembly 106 is configured for capturing skin and dermatoscopy images near-simultaneously, and the illumination assembly 200 is configured for emitting separate light pulses, each with a different spectrum during image co-capture events (e.g., white light for the skin images and NIR light for dermatoscopy images). In certain aspects, the dark period between co-captured images is between 0-5 msec, and the co-captured image center is shifted accordingly.

[0051] In FIG. 3, a flowchart provides a high-level description of an image acquisition process performed using the system shown in FIG. 1. The process involves positioning 300 the subject 103 on the support structure 102. This may involve, among other things, ensuring the subject 103 is within the scanning boundaries of the system 100 and ensuring the subject is oriented to expose the desired surface 105 of the body. At 301, a 3D scan is performed on the subject 103 utilizing the 3D mapping assembly 114 to obtain and process 3D data, e.g., 3D surface mesh, etc.

[0052] At 302, the 3D skin map is used to prepare the scanning plan 412 (FIG. 6A). The scanning plan 412 may include at least one of the following input data for the positioning assembly 104 during the scan: (a) image acquisition assembly 106 X-Y position; (b) image acquisition assembly radial and angular position; (c) adjusting parameter for the focus adjusting unit 218; or (d) change in scan direction. The adjusting parameter provided in step 302 needs to be adjusted due to body motion of the subject 103 and accumulated mechanical drift of one or both of the actuators 108, 110 and in turn the image acquisition assembly 106.

[0053] At 303, the scanning plan 412 is executed, in which a series of overlapping at least partially focused images is taken of the subject 103. At least one sequence of significantly overlapping images can be obtained along the traversal path of the image acquisition assembly 106. The captured photographic images can include medical dermatoscopic information. The at least one sequence of significantly overlapping images can be deblurred using any suitable technique. For example, at 304, the images can be motion deblurred using one or more motion deblur algorithms described herein. In one or more embodiments, each captured image can be motion deblurred utilizing dedicatedmotion compensation software and the relative motion vector data. At 305 the motion- deblurred images can be focus-deblurred using one or more image enhancement algorithms described herein. At 306 the corrected images are stitched into a continuous image (no missing surface sections) of that portion of the subject’s body skin surface 105 using any suitable technique. For purposes of this disclosure, a “continuous image” refers to at least one set of overlapping images obtained by the dermatoscopy system 100 during a complete scan of at least a portion of the subject’s body skin surface 105. An automated processing may be applied to the deblurred image set or the continuous image to identify anomalies, e.g., indicators of a skin disease.

[0054] Note that some operations shown in FIG. 3 may be performed in parallel. For example, if the 3D mapping assembly 114 is fixed or can move independently from the image acquisition assembly 106, then the 3D scan at 301 can be performed over part of the total exposed body area and a partial scanning plan can be developed in 302. The image scanning 303 may begin according to the partial scanning plan while at the same time another part of the exposed body area can be scanned to gather additional 3D data. In this way, at least some portion of 3D scanning and image acquisition can occur in parallel.

[0055] Providing a correct adjusting parameter for the focus adjusting unit 218 requires at least an updated camera - skin FOV center distance at the current image capture timing. One option to provide such adjusted parameter is to utilize the location sensor 115 that provides 5DOF data to the image acquisition assembly 106 relative to the reference plane 120.

[0056] In one or more embodiments, the imaging system 100 may be configured so that the raw captured images of the subject’s skin surface 105 are (a) mutually aligned (as to improve the image stitching process of providing a continuous image); (b) near-focused (as to improve the image deblur process); and (c) captured from a near-nominal distance (as to maintain near-nominal image magnification). One technique for providing these features is to utilize the location sensor 115 that can be configured to provide 5DOF coordinates of the image acquisition assembly 106 relative to the support structure 102. The term aligned 5DOF stands for a 5DOF coordinate system, where one axis is parallel to an image acquisition assembly 106 view axis 107 (FIG. 4A) (hereinafter, view axiscoordinate) and another axis is substantially parallel to the subject’s spinal column (e.g., the Y-axis as shown in FIG. 4A).

[0057] As shown in FIGS. 4A and 4B, the location sensor 115 includes a light detection and ranging (LiDAR) device 452 and a target surface 454. The LiDAR device 452 is fixed to the image acquisition assembly 106 using any suitable technique. Further, the target surface 454 is fixed to the support structure 102 using any suitable technique. A structured printed feature or pattern 462 can be disposed on the target surface 454 (FIG. 4F).

[0058] The LiDAR device 452 can include any suitable devices or components that can be configured to provide at least one of 2D LiDAR, 3D LiDAR, or solid state (nonscanning) 3D LiDAR. The LiDAR device 452 can utilize at least one of the following techniques: Time of flight (ToF), Amplitude Modulated Continuous Wave (AMCW), Frequency Modulated Continuous Wave (FMCW), or other LiDAR technique. The LiDAR device 452 is configured to direct an electromagnetic energy beam along a LiDAR view axis 460 to the target surface 454, where the LiDAR view axis defines an angle theta with the image acquisition assembly view axis 107 of the image acquisition assembly 106 of at least about 30 degrees and no greater than about 90 degrees.

[0059] For typical applications, such as autonomous driving, a LiDAR device FOV can at least partially overlap image acquisition FOV. For example, H. Alismail et. al., “Automatic Calibration of a Range Sensor and Camera System,” Proc. Int. Conf. 3D Imaging (2012), described integrating a LiDAR in a 3D scanning head so that a camera view axis is at a small angle to a LiDAR view axis as to provide focus data for the imaging camera. Camera and LiDAR view vectors were externally aligned using flat patterned targets placed within the scanned volumes and a simple axes alignment routine. They demonstrated a piecewise scan of a room interior without “double walls” (i.e., proper image stitching).

[0060] Automatic scanning of indoor spaces may be conducted without target surfaces. For example, D. Zhang et. al., “An Overlap-free Calibration Method for LiDAR-Camera Platforms based on Environmental Perception,” IEEE Trans. Inst. Meas. V72 (2023), described an indoor scanning device including a LiDAR device fixed to a camera. They further developed an external calibration algorithm for the device to ensure properstitching between the consecutive images captured by the camera. The algorithm corrects the camera view axis to the LiDAR view axis without the need for a target surface. Next, the algorithm places a point cloud on the captured image, i.e., placing each LiDAR point of the cloud on the correct pixel on the captured image. In some embodiments, the image acquisition assembly has a narrow FOV. As a result, the point cloud placement technique described in Zhang may not be as suitable.

[0061] In certain aspects, the LiDAR-based location sensor 115 provides 5DOF coordinates of the image acquisition assembly 106 by capturing the following separate data sets: (a) the image acquisition assembly 106 captures narrow FOV images of the subject’s skin surface 105 at a short working distance; (b) the LiDAR captures a point cloud that covers at least a significant portion of the target surface 454 area, and typically does not cover the subject’s skin surface 105; and (c) the computing unit 122 uses the point cloud data for calculating the LiDAR 5DOF and, in turn, the image acquisition assembly 106 5DOF during the image capture event.

[0062] As mentioned herein, FIG. 4A illustrates the location sensor 115 that includes the LiDAR device 452 fixed to the image acquisition assembly 106, and the target surface 454 fixed to the support structure 102. Rail 109, 111 (not shown) and privacy cover 457 are configured to define an empty volume 456 for providing a clear view path between the LiDAR device 452 and the target surface 454 during scanning plan execution. As shown in FIG. 4A, the LiDAR device 452 view axis 460 may be at a large angle theta (e.g., 90 degrees) relative to the image acquisition assembly 106 imaging axis 107.

[0063] The target surface 454 can include one or more suitable patterns 462, such as at least one of (a) structured printed features; (b) suitable indicia; (c) three-dimensional protruding features; or (d) sub-regions with variable reflectance. The pattern or patterns 462 cover at least a portion of the target surface 454.

[0064] In one or more embodiments, the scanning plan can be executed by (a) capturing a small FOV from the subject’s skin surface 105 during each image capture event utilizing the image acquisition assembly 106, and simultaneously capturing a point cloud from a relatively large FOV including the target surface 454 using the LiDAR device 452; (b) calculating the LiDAR 5DOF relative to the support structure 102 using the point cloud; and (c) calculating the image acquisition assembly 106 5DOF during the imagecapture event, using the LiDAR 5D0F and external calibration data. In certain embodiments, the external calibration may be conducted by sampling the LiDAR point cloud when the image acquisition assembly 106 faces a “home” feature fixed to the support structure 102.

[0065] The points cloud captured by the LiDAR device 452 is sorted to retain the points of target surface 454. Next, the sorted point cloud is processed by the computing unit 122 to calculate the LiDAR device 5DOF with angular accuracy better than 500 microradians. As shown in FIG. 4A, the target surface 454 is substantially perpendicular to the subject’s 103 spinal column. In turn, the LiDAR view axis 460 is substantially perpendicular (theta ~ 90 deg.) to the image acquisition assembly 106 view axis 107. The image acquisition assembly 106 5DOF is calculated from the LiDAR 5DOF. However, the aligned 5DOF coordinate parallel to the view axis 107 is calculated with an error inversely proportional to cos(theta), which might be relatively large.

[0066] In one or more embodiments, the target surface 454 is located adjacent the subject 103 and fixed to the support structure 102. In turn, the LiDAR view axis 460 is tilted at a reduced angle, e.g., 45 degrees relative to the image acquisition assembly 106 imaging axis 107. In that way, the aligned 5DOF coordinate parallel to view axis 107 is accurately calculated.

[0067] Typically, the LiDAR device 452 can capture the point cloud during a short period (e.g., 1 millisecond). During scan plan execution, the image acquisition assembly 106 is continuously moved. In turn, at least some point cloud data may slightly deviate from their true values. An algorithm (e.g. D. Zhang 2023) may use the latest relative speed vector of the image acquisition assembly 106 to compensate for motion-induced points deviation.

[0068] Potential noise in the LiDAR points cloud data may introduce fluctuations into the LiDAR 5DOF history, which can be fdtered. For example, D. Zhang (2023) described a method that applies Kalman filtering on LiDAR-calculated view axis history as to minimize fluctuations. The Kalman filter setting may be determined in consideration of the image acquisition assembly 106 inertia and torque of the actuators 108, 110.

[0069] FIGS. 4C and 4D illustrate another embodiment of a location sensor 115, where the location sensor includes an emitter 430 fixed to the image acquisition assembly 106, atranslucent screen 433 fixed to the support structure 102, and a camera 435 fixed to the support structure 102. The emitter 430 directs a structured electromagnetic energy beam 431 (e.g., light, microwave) to the screen 433. Rails 109, 111 (not shown) and privacy cover 457 are configured to define an empty volume 456 for providing a clear path from the emitter 430 to the screen 433, at least during scanning plan execution. The screen 433 is disposed within the volume 456 and faces the emitter 430 during the scanning plan execution. The angle theta between the emitter view axis 466 and the image acquisition assembly 106 imaging axis 107 is at least 60 degrees and no greater than 90 degrees.

[0070] During scanning plan execution, a distance between the screen 433 and the emitter 430 varies, e.g., between 0.3-2.5 meters. In one or more embodiments, the emitter 430 can be configured to maintain a focused patterned image 434 on the screen 433. For example, the emitter 430 can use a spatial filter 432 and a projection telescope combined with a focus adjustment unit, such as a liquid lens, for projecting a focused pattern 434 on the screen 433.

[0071] In one or more embodiments, the scanning plan can be executed by (a) capturing a small FOV from the subject’s skin surface 105 during each image capture event utilizing the image acquisition assembly 106, and simultaneously capturing the screen 433 image utilizing the camera 435; and (b) processing the captured screen 433 image for calculating the image acquisition assembly 106 5DOF relative to the support structure (FIG. 1).

[0072] In certain aspects, the image acquisition assembly 106 5DOF is calculated by solving a suitable equation set related to projected image 434 parameters. For example, the projected image 434 can be a rectangle and processed for calculating at least one of the following parameters: a) center coordinates; b) image dimensions; c) keystone distortion; or d) transverse keystone distortion. The keystone distortion is proportional to cos(theta). When theta approaches 90 degrees, the keystone distortion approaches zero. In turn, some 5DOF parameters (theta and view axis coordinate) may become inaccurate. In one or more embodiments, the screen 433 is curved. As a result, the induced keystone distortion is maintained within most of the area of the image 434, even for theta approaching 90 degrees. The computing unit 122 can use the enhanced keystone distortion for improving the view axis coordinate accuracy.

[0073] In one or more embodiments, the location sensor 1 15 can be a triangulation distance sensor as described, e.g., in Pigeon et. al., “Using a Slit to Suppress Optical Aberrations in Laser Triangulation Sensors,” Sensors Vol. 24 no. 8 p. 2662, (April 22, 2024). The reference describes an exemplary distance sensor that utilizes a triangulation method. The distance measurement is described as being improved by passing reflected laser light through a diffraction slit. The triangulation distance sensor may require additional data for providing the aligned 5DOF data. In one or more embodiments, the distance sensor readout may be taken simultaneously with at least a fraction of the image acquisition assembly 106 image capture events.

[0074] In certain embodiments, each skin FOV center distance is validated by the distance sensor 450. If the difference between the location sensor 115 calculated distance and the measured distance from the distance sensor 450 exceeds a selected threshold, the scanning plan can be updated accordingly, using any suitable technique.

[0075] FIGS. 10A and 10B illustrate another example of a location sensor 715 firmly fixed to the image acquisition assembly 106 of system 100. The location sensor 715 in this example includes an emitter 700 that sends out an energy beam 701 structured by a filter 702 (e.g., light, microwave, ultrasound) to a suitable screen 703 positioned 0.3 - 1.5 meter from emitter 700. The illuminated screen 703 is imaged by an imaging camera 705 with a detector 704 and both are fixed in respect to the support bed 102. The camera 705 captures the screen 703 image simultaneously with each camera 212 capture event. A processing unit 708 processes the captured screen 703 image and explicitly calculates the dermatoscopy camera 212 optical axis vector in respect to the support bed 102 at estimated accuracy of 500 microRadians. The crossing point between that vector on the stored 3D subject’s body topography surface is the XYZ coordinates of the FOV center of the latest captured image.

[0076] During execution of the image scan plan, the displacement system 108 may accumulate drift so that image system 106 location and orientation deviates from its planned location / orientation according to the scan plan. The processing unit 708 corrects at least a fraction of that drift using the location sensor 715 readout. A complementary correction may use a distance sensor that measures the distance between the camera 212 and the skin.

[0077] In FIG. 4B, a diagram illustrates an example of a distance sensor 750 operable for drift correction. The distance sensor 750 is fixed to the image acquisition assembly 106. In certain embodiments, the distance sensor 750 sends an energy beam 751 towards a point on the subject body skin 752. The skin reflected beam energy is imaged and received on a detector array 753 fixed in respect to the image acquisition assembly 106. The imaged skin spot location on the detector 753 is converted into a signal coupled to a processor 754. The distance sensor is installed on the image system 106 so that beam 751 coincide with the image camera 212 axis on the subject body skin. In certain embodiments, the distance sensor readout is taken simultaneously with certain camera 212 image capture events. The drift can be calculated by comparing the distance sensor 750 output converted into skin point XYZ with the respective skin point XYZ calculated from the location sensor 115 output.

[0078] Each skin point distance measured by the distance sensor 750 is compared to the respective skin point distance calculated from the updated image scan plan. A dedicated algorithm is used to convert each above threshold / systematic difference between the two values for correcting the image scan plan. The scan plan correction may result with liquid lens adjustment instructions as necessary to achieve best imaging focus.

[0079] In one or more embodiments, the image acquisition assembly 106 can capture an image sequence of at least a portion of the subject’s body surface at high rate in on-the- fly mode. In certain aspects the capture rate is sufficient to enable capture of small FOV images at significant overlap between sequentially captured images. Eliminating the need to stabilize the image acquisition assembly 106 before each image capture significantly reduces the subject’s body scan duration. However, operating at the on-the-fly mode potentially induces motion blur in the captured images for a lighting unit 200 emitting light at a finite (e.g., 1 msec) pulse width.

[0080] While not wishing to be bound by any particular theory, the image captured during motion can be expressed as the sum of a blur kernel convolved original image plus noise term. The image deblur process includes the estimation of the blur kernel and deconvolving the original object’s image. A convolutional neural network (CNN) algorithm is a common technique for image deblur and in particular video deblur. For example, Zhan, et. al., “Video deblurring via motion compensation and adaptive informationfusion,” Neurocomputing Vol. 34 no 1 pp. 88-98 , 2019, demonstrated image deblur and 3D scene reconstruction from a video clip captured by a non-stabilized camera.

[0081] In certain embodiments, the captured images’ motion induced blur is corrected first, followed by 3D induced blur correction. In certain aspects, the current and latest relative motion velocity vectors are near-parallel, and the relative speed is near-constant. In such case, the blur kernel can be separated into a motion blur kernel and a focus kernel. For example, see Z. Zhang, “Blind Remote Sensing Image Deblurring Based on Overlapped Patches’ Non-Linear Prior,” Sensors Vol. 22 p. 7858. (2022).

[0082] In certain aspects, the motion blur induced at captured photographic image of a 3D object may be non-uniform. See Tran, et. al., “Explore Image Deblurring via Encoded Blur Kernel Space,” CVPR 2021 Conf, paper. The non-uniform motion blur can be corrected utilizing any suitable technique. In one or more embodiments, the motion blur can be corrected by a) updating the relative motion vector, using at least two latest 5DOF data points and the stored subject’s skin map; b) generating a motion blur kernel using the relative motion vector and the updated FOV depth map; and c) using the motion blur kernel for correcting the captured raw images.

[0083] In other aspects the current and latest relative motion vectors are not parallel (e.g., chin to neck transition), In such case the motion blur kernel can be calculated using a modified relative motion vector, e.g., using a suitable linear interpolation between the current and latest relative motion vectors.

[0084] During scanning plan execution, the subject’s skin surface 105 within the image acquisition assembly 106 FOV can deviate from the skin FOV. In such cases, the focus adjustment unit 218 setting may require optimization utilizing one of the following steps: (a) measuring the distance to the FOV surface using a distance sensor; (b) estimating the distance using the image acquisition assembly 5DOF coordinates calculated from the location sensor 115 readout and the stored 3D map data; or (c) estimating a distance compensation value from data obtained during the deblur process of the previous captured image. The computing unit 122 may send an incremental signal to the focus adjustment unit 218 based on this optimized distance value.

[0085] Another approach to correct 3D induced blur is to utilize a telescope 216 with high optical aperture setting. However, the optical aperture value is determined by atradeoff between a) image acquisition assembly 106 light collection efficiency; and (b) depth of focus (DOF) required for sharp FOV imaging. For example, in certain embodiments, the telescope 216 optical aperture is f / # = 4.0. Increasing the optical aperture to f / # = 8.0 would enable DOF = 150 microns (ideal resolution = 4 microns) while reducing the light collection efficiency to 25% of the collection efficiency at the original telescope setting. Some skin regions 105 of the subject 103 can be characterized with a valid FOV depth map 343 whose amplitude exceeds by far 150 microns.

[0086] The 3D blur may be corrected using any suitable technique. In one or more embodiments, the 3D blur can be corrected by utilizing one or more of the following techniques: a) focusing using the focus adjustment unit 218 and the 5DOF data estimated from the location sensor 115 readout; b) utilizing a telescope 216 with high optical aperture setting; c) using a sequence of images and a suitable algorithm for reconstructing a deblurred image; or d) deblurring using properties of suitable metalens and a suitable CNN algorithm.

[0087] The telescope 216 optical aperture value is determined by a tradeoff between a) image acquisition assembly 106 light collection efficiency; and (b) depth of focus (DOF) required for sharp FOV imaging. For example, in certain embodiments, the telescope 216 optical aperture is f / # = 4.0. Increasing the optical aperture to f / # = 8.0 would increase the DOF to 150 microns while reducing the light collection efficiency to 25% of the collection efficiency at f / # = 4.0 setting. However, some subject’s skin surface 105 FOV may include regions whose amplitude exceeds 150 microns.

[0088] In certain embodiments, the motion-deblurred image is focus-deblurred using a CNN algorithm. Typically, the CNN focus deblur process involves the following steps: a) calculate the image acquisition assembly 106 focus surface ("current focus surface") at the current image capture; b) calculate the image acquisition assembly 106 view axis (current view axis) using the location sensor 115 readout sampled at the current image capture; c) calculate the estimated depth (out-of-focus) 3D map, using the stored 3D map, the current view axis and the current focus surface; d) calculate the estimated blur kernel from the estimated 3D depth map; e) deblur the current image using a proper CNN algorithm, an estimated weights matrix and the estimated blur kernel; and f) correcting artifacts generated during de-convolving the current image with the optimized blurkernel .

[0089] In certain aspects, the CNN focus deblur algorithm calculates a blur map after each de-convolving cycle, using a proper blur estimator. In certain aspects, the blur estimator output is used to divide the processed image into regions, each with different blur amount categories, e.g., bad blur, median blur, marginal blur, and no-blur. The weights matrix is optimized at each cycle until most image regions are categorized at the no-blur level. In certain aspects, the convergence may be improved by comparing the blur amount map to the estimated 3D depth map.

[0090] One or more weights matrices suitable for quick convergence of the focus-deblur CNN algorithm can be estimated, prior to a scanning plan execution, through suitable training process. The training process may be conducted externally and may use multiple sets of same skin FOV, including highly focused images and captured images (with known blur level) and estimated 3D depth map.

[0091] The use of the estimated 3D depth map significantly shortens the image focus deblur processing time using a CNN algorithm. For example, Anwar et. al., “Depth Estimation and Blur Removal from a Single Out-of-focus Image,” BMVC conf. Vol. 1 p. 2, 2017, demonstrated correction of various 3D blurred still images when providing their respective estimated 3D FOV depth map. The use of the estimated weights matrix, as described herein, can further accelerate the CNN algorithm convergence. Specific estimated weights matrix may be used for focus-deblur / moti on-deblurred images captured from transition regions such as the chin-neck region.

[0092] In certain aspects, the CNN algorithm, after converging, may calculate an improved 3D Depth map, using the final blur kernel, e.g., as described by Anwar (2017). The improved 3D depth map generation may be assisted with overlapping regions from at least one previous depth map. In certain aspects, certain improved 3D depth maps may be compared to the estimated depth map to detect depth map deviation beyond certain boundaries.

[0093] As used herein, the term “valid 3D FOV depth map” refers to the difference between the actual skin 3D surface and the aligned skin FOV at that image capture event. As used herein, “view pyramid” refers to a pyramid whose apex is along the telescope axis and its base is the aligned skin FOV. As used herein, the term “morphing kernel”refers to a matrix that morphs a FOV image captured by the image acquisition assembly 106 into a photogrametric image of that FOV. In certain locations during execution of the scanning plan 412 (e.g., at transitions from face to neck region), the first estimated FOV depth map is calculated from the difference between the estimated 3D surface map and the updated skin FOV.

[0094] A CNN algorithm that co-generates an updated depth map would share overlapping regions from at least one previous depth map. In certain aspects, each captured image would be accompanied with a depth map improved by using a CNN algorithm configured for co-generating a FOV depth map.

[0095] In certain aspects, there is a need to refresh the FOV depth map at one or more of the following events: (a) starting the subject’s body scanning procedure; (b) at a beginning of a new image capture sequence; (c) at a transition between surfaces (e.g., from face to neck); or (d) following detection of deviation between the XYZ coordinates calculated from the location sensor 115 readout, and the current scanning plan 412 XYZ coordinates.

[0096] Due to the short working distance, the captured skin image is not photogrametric (the X-Y coordinates of a point on the captured image may not be directly converted into the actual skin point X-Y-Z coordinates). The refreshed FOV depth map may be prepared from the 3D stored map utilizing any suitable technique, e.g., (a) cutting a section around the optical axis 107 (FIG. 4B) intersection with the 3D stored map, sufficiently larger than current image acquisition assembly 106 FOV ; (b) calculating a morphing kernel using the cut section and the current working distance; and c) converting the cut section into the captured image format, using the morphing kernel.

[0097] “FOV focus surface” refers to the focal plane of image acquisition assembly 106 camera at the current focus adjustment unit 218 setting. Herein an aligned focus surface 472 (FIG. 4E) refers to the FOV focus surface of the image acquisition assembly 106 positioned at a known 5DOF. Herein, the term “mapped FOV surface” 473 represents the view pyramid intersection with the stored 3D skin map obtained by the 3D mapping assembly 114. The term “estimated FOV depth map” refers to the FOV depth map available as input to the current captured image deblur process. The term “valid FOV depth map” 474 is the depth map of the actual skin surface within the FOV, relative to thealigned FOV focus surface 472 of that image capture event. Fig. 4E illustrates the various surfaces used for the above mentioned depth maps.

[0098] In Fig. 4E the surfaces involved in calibrating the estimated FOV depth map are described. The image acquisition assembly 106 is characterized by a FOV and near-flat focus surface. At the current 5DOF position, and telescope 216 focusing unit 218 setting, the focus surface is termed aligned focus surface 472, whose extent is defined by the FOV or the base of the view pyramid 471. The aligned focus surface 472 may be perpendicular to the telescope optical axis 217.

[0099] As used herein, the term “valid 3D FOV depth map” refers to the difference between the actual subject’s skin 3D FOV surface 473 and the aligned focus plane 472, at that image capture event. In one or more embodiments, the image captured by the image acquisition assembly 106 is deblurred using a suitable algorithm and an estimated 3D depth map. In certain embodiments, the estimated FOV depth map may be calculated by: (a) cutting a raw depth map whose size is at least the intersection between the view pyramid and the current region within the subject’s body map, prepared by the 3D mapping assembly 114; (b) subtracting the aligned focus plane from the mapped surface FOV; and (c) improving the raw depth map with any suitable data (e.g., previous estimated FOV depth map). In certain embodiments, the estimated FOV depth map converges towards the valid 3D FOV depth map after a few image capture events.

[0100] In certain aspects, the subject’s body surface map, prepared by the 3D mapping assembly 114, is stored as a photogrametric map. The 3D subject’s surface topography map induces distortion of the captured image map grid vs. the stored map grid. For example, a protruding region on the subject’s skin surface 105 (e.g., ankle) may be viewed relatively larger in the captured image vs. within the photogrametric 3D map. In certain aspects, the stored surface map is converted into the estimated depth map by: (a) cutting a raw depth map whose size is at least the intersection between the view pyramid and the subj ect’ s body surface map; (b) morphing the photogrametric map section by converting its grid into the captured image grid; (c) subtracting the aligned focus plane from the morphed surface FOV map; and (d) improving the raw depth map with any suitable data (e.g., previous estimated FOV depth map).

[0101] For relatively flat subject’s skin regions, the current FOV depth map maybe calculated by updating one or more previous FOV depth maps alone. However, preparing FOV depth maps for fast varying skin topography (e.g., chin to neck transition) may involve all three steps above.

[0102] In certain embodiments, the raw images are captured using imaging optics including the telescope 216 and a metalens. These raw images can be deblurred using any suitable technique, e.g., a CNN algorithm and an updated FOV depth map. Various metalens types and suitable algorithm schemes may be employed for correcting the captured image deblur as is further described herein.

[0103] In certain embodiments, the metalens has a high chromatic dispersion, and the skin image sensor 222 can include multiple chromatic fdters , e.g., RGB. The image deblur can be conducted through a CNN algorithm that utilizes an initial 3D depth map (e.g., updated 3D depth map). For example, Lach et. al., “3D Imaging Using Extreme Dispersion in Optical Metasurfaces,” ACS Photonics vol. 8 no. 5 pp. 1421 - 1429, 2021, fabricated a metalens with extreme chromatic dispersion and reconstructed the RGB image and depth map of a 3D object in a single shot. Further, a CNN algorithm for image deblur was developed. First, a 3D reference colored object was captured with an RGB camera. Next, the CNN algorithm was trained on the three channels (RGB) of the reference object image. A new colored 3D object was captured, and its image was separated into its RGB channels. The trained CNN algorithm was applied to the new captured image channels, and a high resolution RGB image and a FOV depth map were constructed. The constructed FOV depth map was near-identical to the actual FOV depth map and sufficient for complete captured image deblur.

[0104] In certain embodiments, the skin and dermatoscopy images, captured from the same skin FOV, are deblurred according to the following steps: a) motion deblurring the skin image using a suitable motion kernel; b) motion deblurring the dermatoscopy image using the same motion kernel; c) focus deblurring the skin image using suitable CNN algorithm, estimated depth map, and estimated weights matrix; d) calculate the improved 3D depth map after converging the deblur process; and e) focus deblurring the dermatoscopy image using the same improved 3D depth map.

[0105] In certain embodiments, at least one sequence of images captured from a subject’s skin region is first deblurred and then stitched to a larger continuous image.Proper image stitching may be conducted per the following stages: a) overlapping images registration; b) rotational alignment between overlapping images c) adjust magnification of overlapping images regions; d) stitch images through iterative process, according to a set of criteria; and e) minimize artifacts within the stitch region. For example, Yoon et. al., “Real-Time Video Stitching Using Camera Path Estimation and Homography Refinement, “ Symmetry vol. 10, no. 4, 2018, demonstrated a continuous and artifact-free image stitching from some overlapping raw images captured along a scan path. Continuous (piecewise) path estimation combined with homography refinement for properly stitching the images into a continuous image was utilized.

[0106] The registration of the stitched images may be calculated from view axis 5DOF, calculated from the location sensor 115 readout. The linear image magnification at the overlap region may be calculated from the 3D depth map and the focus plane data of each stitched images. However, the image alignment may require image rotation matrix optimization through an iterative process, e.g., based on homography refinement as described, e.g., in S. Wen et. al., “Investigation of image stitching refinement with enhanced correlation coefficient,” Malaysian J. Comp. Sci, p. 22 - 34 (2020). They developed a fast wavelet-based algorithm that estimated the likelihood of small sections within the overlap region common to the stitched images. They demonstrated stitching of multiple highly misaligned images with no straight lines into a continuous piecewise panoramic image. In other aspects, the rotation matrix may be estimated from previous rotation coordinate values.

[0107] In certain embodiments, the iterative alignment process is conducted first on the skin stitched images, and the optimized rotation matrix is calculated. Next, the dermatoscopy images stitching is conducted according to the following steps: a) registration of the stitched images; b) application of the optimized rotation matrix on both images; c) application of the magnification vector for each image; d) minimization of artifacts from the stitched region.

[0108] FIG. 5 describes one embodiment of an image deblur and stitching process flow of the captured images. At step 535, the image acquisition assembly 106 captures near-simultaneously two photographic images that can include a skin image and a dermatoscopic image. The dermatoscopy imaging path includes the polarizer 220 whosepolarization plane is perpendicular to the illumination path polarizer 210.

[0109] The directly reflected polarized photons are mostly blocked by the polarizer 220 within the dermatoscopy imaging path. In turn, each dermatoscopy captured image represents mostly in-skin (back-scattered) propagating photons whose flux is much lower than the directly reflected photons. Further, imaging the back scattered photons lacks most small skin surface details. In contrast, image deblur processes depend on the high frequency image components (i.e., small features). This tradeoff supports a deblur process, where the skin image is deblurred first, and the deblur by-products (e.g., motion kernel, depth maps) are used to deblur the dermatoscopy image.

[0110] At step 540, the captured skin image is motion deblurred using a motion deblur kernel. First, the relative motion vector is estimated from the scanning plan 412 and the recent location sensor 115 readout 5DOF coordinates. Next, the initial motion deblur kernel is calculated using the motion vector. Next the motion deblur kernel may be morphed using the updated depth map.

[0111] The captured skin image is motion deblurred utilizing the motion deblur process that may be conducted through one or more cycles. For example, the motion deblur kernel may be updated so as to achieve near-uniform high frequency content throughout the motion deblurred skin image.

[0112] In parallel, the captured dermatoscopic image is motion deblurred utilizing the updated motion deblur kernel.

[0113] At step 545, the motion-deblurred image is focus deblurred through a CNN architecture and two 3D depth maps. An example of this deblur process is given in Anwar et. al., “Deblur and deep depth from single defocused image,” J. Meeh. Computer Vision vol. 32, art. 34, 2021. First, two 3D depth maps are prepared. The first is the latest updated depth map and the second map may be the first depth map with modifications.

[0114] Next, the CNN architecture generates two deblurred skin images utilizing each of the depth maps. Some regions in the second deblurred image may exhibit different deblur compared to the those of the first deblurred image. Next, the CNN algorithm blends the two deblurred images so as to obtain an enhanced-resolution image. The CNN algorithm decides which regions will be taken from each deblurred image into the blended image. The CNN algorithm may be trained from blending of previousdeblurred images prepared from previous overlapping captured image, and in turn, can accelerate the processing time and improve deblur quality.

[0115] Next, skin images can undergo noise filtering and edge enhancement processes that remove noise and tune the local edges intensity towards typical levels of focused skin images using techniques known in the art.

[0116] In parallel at 540 and 545 of FIG. 5, the CNN architecture utilizes the first and modified depth maps for generating the two deblurred dermatoscopy images. The CNN algorithm blends images into a blended image. Next, the blended dermatoscopic images undergo noise filtering and edges enhancement processes, equivalent to the respective skin image process.

[0117] At step 550, a sequential set of deblurred skin images are combined into a super-resolution skin image utilizing an MFSR algorithm. MFSR refers to the process of estimating a high-resolution image from a sequence of lower resolution observations as described, in Ma, “Handling Motion Blur in Multi-Frame Super-Resolution Computer Vision and Patterned Recognition,” 2015 IEEE Conference on Computer Vision and Patterned Recognition (CVPR, Boston, MA pp. 5224- 5232, 2015). This reference describes a three-fold MFSR process. First, the motion blur was analyzed with feedback control (active image registration). Second, informative structures were extracted from each frame. Third, a significant portion of the noise was suppressed using a spatial sparsity technique. Briefly, the sparsity technique effectively extracted the outline of each image. Next a sharp image was reconstructed (with minimal noise) based on the common outlines found.

[0118] At least three sequential blended skin images with overlapping regions participate in generating a high-resolution skin image whose FOV may be identical to each of single participating blended skin images. During the MFSR process, one or more images may be properly aligned towards a common axis. Using 3 MPixel raw skin images, it is expected that at least a portion of the super-resolution image will have a near-micron image plane resolution.

[0119] In parallel, a sequential set of blended dermatoscopic images may be processed by an MFSR algorithm into an equivalent sequence of high-resolution dermatoscopy images.

[0120] At step 555, each super-resolution skin image is morphed into a photogram etric format skin image utilizing the de-morphing kernel. The de-morphing kernel is calculated from the updated FOV depth map.

[0121] In parallel, the super-resolution dermatoscopy image is morphed into a photogram etric format dermatoscopy image using the same de-morphing kernel. De- morphing images into photogrametric format images simplifies various processes such as image stitching. At step 560, a sequence of photogrametric format skin images are stitched into a continuous skin image of at least a portion of the subject’s body. In parallel, a sequence of photogrametric format dermatoscopy images are stitched into a continuous dermatoscopy image of at least a portion of the subject’s body. The stitching instructions (i.e., overlapping regions removal) may be similar to those used to stitch the skin image.

[0122] In certain aspects, the skin images are stitched into a continuous image such that each X-Y point on the continuous skin image can be directly related to a respective X-Y-Z point on the subjects’ body surface. Arranging the continuous skin image in such format may be advantageous for a) proper detection of suspicious skin structures divided between two consecutive captured images; and (b) repeated capture of an incriminated skin region for validating the finding.

[0123] The dermatoscopy images may be stitched into a continuous dermatoscopy image with direct relation between an X-Y point on the continuous dermatoscopy image and the respective X-Y-Z point on the subject’s body surface. A continuous dermatoscopy image with direct coordinates relationship may be advantageous for incriminating a skin region based on findings from the continuous images.

[0124] In certain embodiments, the raw image sequences are converted to continuous images, adhering to the photogrametric format, using any suitable technique. In one or more embodiments, the raw image sequences can be converted to continuous images by (a) extracting a FOV depth map section from the stored 3D body map using the 5DOF coordinates and the view pyramid 471; b) calculating a morphing kernel using the FOV depth map; c) morphing the raw skin and dermatoscopy images into photogrametric format images; d) correcting the motion blur of the morphed skin anddermatoscopy image pair using the relative motion vector; e) deblurring the corrected skin image using a suitable CNN de-blur algorithm and the estimated FOV depth map and generate an improved FOV depth map; f) deblurring the corrected dermatoscopy image using the CNN algorithm and the improved FOV depth map; g) enhancing the skin and dermatoscopy de-blurred image pair using an MF SR algorithm and the appropriate de-blurred image sequence; and h) stitching sequences of enhanced skin and dermatoscopy images and improved 3D FOV depth maps into a continuous skin and dermatoscopy images and continuous 3D depth map.

[0125] In one or more embodiments, certain estimated FOV maps (e.g., at face to neck transition) are converged towards the respective valid 3D FOV depth map by again deblurring the corrected skin image using a suitable CNN deblur algorithm and the estimated FOV depth map and generating an improved FOV map. That FOV depth map can assist proper deblurring of skin images covering each transition region (including the first captured image pair).

[0126] As shown in FIG. 6A the scanning plan 412 mentioned herein includes a data array (e.g., a camera triggering timing) to take overlapping images 502, a subset of which are shown in FIG. 6A. This includes beginning image 502a and ending image 502b according to their order on the traversal path 500. FIG. 6B is a diagram that illustrates how support 112 can be articulated (e.g., extended and rotated) to allow the image acquisition assembly 106 to be separated from the subject 103 by a predetermined working distance 600 (within some allowable tolerance) during the traversal. A “working distance” refers to the distance of the distal camera telescope 216 surface from the center of FOV surface 434. In one or more embodiments, the working distance can be no greater than 30 cm. In one or more embodiments, the nominal working distance is at least about 50 mm and no greater than about 200 mm. An optimum traversal path 500 can be found using any suitable algorithm, for example, one or more embodiments of algorithms described in Gasparetto et. al., “Path Planning and Trajectory Planning Algorithms: A General Overview,” Motion and Operation Planning of Robotics System: Background and Practical Approaches, vol. .29, pp.3-27 Mechanisms and Machine Science, 29, Springer 2015.

[0127] FIG. 6B illustrates a sequence from time tl to time t3 in which a linearactuator 601 of the support 112 is extended / retracted, and a joint 602 is rotated while the actuator moves the support 112 and image acquisition assembly 106 along the rail 111. The joint 602 may be considered an actuator in this context. This allows not only maintaining the working distance 600 within tolerance, but also allows the image acquisition assembly 106 to be aligned with a near-normal vector of the surface being imaged. Aligning the image acquisition assembly 106 has advantages in terms of (a) focusing (e.g., keeping a larger fraction of the captured image focused) and (b) lighting uniformity (e.g., near constant illumination flux). This simplified example shows rotation about one axis (Y-axis), but the joint 602 may rotate about two axes, providing a full five DOFs (5DOF) positioning and orientation of the imaging system.

[0128] The image scanning plan 412 may also include lighting parameters modifications that optimize the camera images based on the subject body region being imaged. The image scanning plan 412 may be presented as a timed array of signals used to control the image acquisition assembly 106 position and orientation (e.g., utilizing actuators 108, 110, support 112). A simplified example is shown in the graph set illustrated in FIG. 7. The graph set is shown as dependent on time along the horizontal axis, although the scanning plan 412 itself may not be time dependent. For example, the scanning plan 412 may specify a sequence of simultaneous XYZ coordinates and angles 01, 92 of the image acquisition assembly 106 without any specific references to time. The scanning plan 412 should be built to minimize image acquisition assembly 106 acceleration / deceleration as to minimize its vibration imposed on its motion according to the scanning plan 412.

[0129] The illustrated motion plan includes a series of passes in which the camera shutter is opened and / or lighting system flashes is repeatedly actuated, and turn-around regions 604 in which, for example, the actuators are preparing for the next pass and the image acquisition assembly 106 is outside the boundary of the subject and is therefore not taking images. The motion plan may seek to minimize these turn-around regions to reduce acquisition time, e.g., by using a spiraling path from an outer boundary inward instead of linear passes.

[0130] The camera triggering and illumination assembly inputs shown in FIG. 7 are one example of signals input to the image acquisition assembly 106. Other imageacquisition inputs may include focusing values input to the focusing unit 218 (indicated by the trace in the graph), lighting parameters sent to illumination assembly 200, parameters for the visible and dermatoscopic digital image sensor(s), etc. Generally, these parameters are derived based on the 3D scanning plan 412 and the subject body region being scanned.

[0131] In certain embodiments, the lighting parameters may be modulated while capturing dermatoscopic images. For example, processing a pair of skin images captured with alternating illumination spectrum may enhance white streaks in the imaged pigmented nevi which are an indicator for malignant nevi.

[0132] In FIG. 8, a block diagram shows some functionality of the computing unit 122 according to an example embodiment. The computing unit 122 may include one or more computing arrangements with processing hardware such as central processing units (CPUs), sub-processors (e.g., digital signal processors or DSPs), volatile memory (e.g., dynamic random-access memory or DRAM), persistent memory (e.g., solid-state drives, hard disk drives), input / output (VO) busses, and the like. The VO busses facilitate communication with input devices such as the 3D sensor(s) 802 (e.g., 3D mapping assembly 114), light sensor 806, position sensor 804 (e.g., location sensor 115), stored database tables and subject interface inputs 808 (e.g., mouse, keyboard, touch screen, biometrics), etc. The I / O busses also facilitate communication with output devices such as imaging devices and optics 810 (e.g., imagers, focusing assembly, illumination assembly 200, polarizers), positioning devices 812 (e.g., step motors, linear motors, servo motors), and subject user interface 814 (e.g., monitors, speakers).

[0133] The computing unit 122 has a topography generation module that processes the data acquired by the topography generation unit 820. This data is processed to form a frozen body 3D map 821 of at least a portion of the subject body surface 105. The frozen body 3D map 821 includes surface points whose XYZ coordinates are identical to users’ skin points coordinates during the scan, relative to the reference plane 120. The generation module also corrects the scan plan 823 according to location sensor 115 data and occasionally by the distance sensor 450 output.

[0134] The user's natural breathing can introduce cyclic deviation from the 3D topography map 821, in particular in the abdominal skin region. In certain embodiments,the 3D mapping assembly 114 generates several 3D topography maps 821, along with their timing. The topography generation module 820 builds a cyclic variation 3D map representing the subject’s breathing cycle, on top of the 3D topography map 821. Various means can be used for compensating the subject’s breathing induced 3D topography 821 deviation. In certain embodiments, the mapping system uses the location sensor 115 readout in combination with the variation 3D map to continuously update the transverse path plan before the next image capture. The updated transverse plan parameters typically include the camera Z-axis position and liquid lens 218 focusing data.

[0135] In certain aspects, a scanning plan 823 generation module 822 also corrects a scanning plan 823 according to location sensor 115 5DOF data and occasionally by the distance sensor 804 output.

[0136] The scanning plan 823 generation module 822 uses the 3D frozen body map 821 to generate the scanning plan 823 that is used by the positioning device 812 that controls an image acquisition module 824 position and orientation during executing the scanning plan 823. The scanning plan 823 includes a trigger sequence signal for the skin and dermatoscopy imaging devices 810, as well as other signals. The scanning plan 823 generation module 822 makes reference to camera data 825 to include a triggering plan of camera input signals 830 sent among other to the dermatoscopic image sensor 214 and skin imaging sensor 222. Other data used for controlling output devices (e.g., illumination assembly 200, positioning) can also have an analogous data set accessible by the computing unit 122.

[0137] Per the scanning plan 823, the image acquisition module 824 is moved by the positioning device 812. During the scan, the image acquisition module 824 captures a sequence of overlapping dermatoscopy and skin images 827 per the trigger sequence signal. At each image capture event, the computing unit 122 triggers the illumination assembly 200 and the location sensor 115 . The computing unit 122 receives the location sensor 115 readout and converts it into the image acquisition module 824 5DOF which is stored together with the captured raw images 827.

[0138] At start and at each transition event (e.g., from chin to neck), the computing unit 122 refreshes the estimated 3D depth map using a section cut from the 3D frozen body map 821, per the current image acquisition module 824 5DOF data, by (a)cutting a 3D map section sufficiently larger than current FOV section from the 3D body map 821; b) calculating a morphing kernel using the cut section 3D map, the image acquisition assembly 106 5DOF data 829 and the focus surface 472; and c) converting the 3D map section into an estimated FOV depth map, compatible with the captured image, using the morphing kernel.

[0139] The image acquisition module 824 provides sets of overlapping dermatoscopy and skin image pairs 827. The computing unit 122 corrects the motion blur of each raw skin image and dermatoscopy image pair, by (a) calculating an estimated motion kernel using current estimated FOV depth map and current relative motion vector; and b) deblurring each raw skin and dermatoscopy image 827 by processing them with the current motion kernel.

[0140] The image post processing module 828 deblurs each motion-deblurred skin image by (a) focus deblur using suitable algorithm (e.g., CNN algorithm); and (b) image enhancement using the present deblurred image, a set of processed skin images, utilizing a suitable MFSR algorithm.

[0141] The image post processing module 828 deblur each motion-deblurred dermatoscopy image by (a) focus deblur using suitable algorithm (e.g., CNN algorithm and data generated during deblurring the paired skin image; and (b) image enhancement using the present deblurred dermatoscopy image, a set of processed dermatoscopy images and a suitable MFSR algorithm. Between transition events, the image post processing module 828 continuously improves the FOV depth map using data generated during the captured images 827 deblur process.

[0142] The image post processing module 828 is configured to stitch a set of processed skin images into a continuous high resolution skin image by (a) calculating the reverse 3D kernel, for each processed skin image, using the improved 3D FOV depth map; b) converting each processed skin image into an equivalent photogrametric image using the said reverse kernel; and c) stitching the sequence of photogrametric skin images into a continuous high resolution skin image.

[0143] The image post processing module 824 stitches sets of high resolution dermatoscopy images into a continuous high resolution dermatoscopy image by (a) converting each de-blurred dermatoscopy image into an equivalent photogrametricdermatoscopy image using the reverse kernel of the paired skin image; and (b) stitching the sequence of photogrametric dermatoscopy images into a continuous high resolution depth map. The image post processing module 828 is further configured to stitch a sequence of FOV depth maps by (a) converting each improved FOV depth map into an equivalent photogrametric depth map using the reverse kernel of the paired skin image; and (b) stitching the sequence of photogrametric depth maps into a continuous high resolution depth map. In one or more embodiments, the skin images are deblurred first, and the dermatoscopy images are deblurred utilizing data generated during the skin image deblur process, where the deblurred skin images sequence is stitched first and the dermatoscopy images sequence is stitched utilizing data generated during the respective skin images stitching.

[0144] The high resolution continuous dermatoscopy, skin and depth map images are suitable for manual or automated analysis. In one or more embodiments, the dermatoscopy continuous image may be used for detecting anomalous under-skin images (e.g., suspected skin lesions). In one or more embodiments, the skin continuous image may be used to detect skin structures (e.g., actinic keratosis, moles, skin marks, etc.) as well as identifying healthy skin. The subject interface input 808 and subject interface output devices 814 are shown facilitating access to the final images, although these subject interface components may be used to control, monitor, and configure any of the functional modules in the computing unit 122.

[0145] In some embodiments, data sent from the location sensor 802 is recorded simultaneously with each image capture event for estimating the five DOF data 829, which is recorded with each image 827 captured by the image acquisition assembly 824. The acquired image stream 827 and five DOF data 829 are combined into high resolution frames partially or fully covering the user body surface. The five DOF data 829 can be compared with the calculated distance from scanning plan 823 so as to identify and correct possible drift between scanning plan versus the executed scan.

[0146] The systems and methods described herein are used for checking skin condition in the scan on-the fly mode and enable fast high-resolution imaging of at least a portion of the subject body skin surface. In this mode, the positioning devices 812 do not stop during the scan. Typical transverse scan speed may range from 5 to 20 cm / sec. Incertain embodiments, the positioning device is configured to drive the image acquisition assembly 106 to a specified 5DOF position and re-capture images of interest.

[0147] With reference to FIG. 9, a flowchart illustrates an example embodiment of processing steps executed by the computing unit 122. This method is initiated once a subject 103 to be examined is accommodated within the support structure 102 of the dermatoscopy system 100, and further when the image acquisition assembly 106 is ready to move towards a first body part (e.g., the torso) of a given body side (e.g., the ventral side of the body) of the subject 103.

[0148] At step 901, the computing unit 122 is configured to activate a displacement element (e.g., actuators 108, 110) to move the 3D mapping assembly 114 and acquire 3D maps that are tailored to at least a portion of the body side of the subject 103. At step 902, the computing unit 122 is configured to analyze the acquired 3D images and prepare a scanning plan 412 for the image acquisition assembly 106 of the first portion of exposed body side of the subject 103. The scanning plan 412 considerations can include at least one of the following: a) near-preserving the working distance during the scan; b) minimizing speed variations during track reversal and angular adjustments; or c) preserving near-nominal captured image overlapping during scan.

[0149] At step 903, the computing unit 122 is configured to execute the scanning plan 412 that includes the motion plan to automatically activate the displacement element (e.g., actuators 108, 110 and support 112). The scanning plan 412 includes an imaging plan with instructions to the image acquisition assembly 106 to acquire a sequence of simultaneous raw dermatoscopy and skin images. The sequence is recorded while the image acquisition assembly 106 hovers at near-working distance, e.g., at approximately 50-150 mm. During executing the scanning plan 412, the illumination assembly 200 can be utilized to simultaneously illuminate the subject while significantly overlapping photographic images of the subject are captured. The captured photographic images include captured dermatoscopy images and captured skin images. In one or more embodiments, the dermatoscopy and skin images are captured near-simultaneously as separate images. In one or more embodiments, the illumination assembly 200 can be configured to emit selected spectra and intensities for each dermatoscopy image and for each skin image. Further a location sensor readout from the location sensor 115 can beconverted into acquisition assembly 5D0F coordinates for each captured photographic image using any suitable technique.

[0150] At step 904, the computing unit 122 is configured to calculate the initial and estimated FOV depth map by (a) calculating the initial 3D depth map using the image acquisition assembly 106 5DOF, the view pyramid 471, and the stored 3D surface map; (b) calculating an estimated morphing kernel from the estimated FOV depth map; and (c) using the initial morphing kernel for morphing the initial 3D depth map into the estimated FOV depth map. The estimated FOV 3D depth map may be improved at some of the captured events, as explained herein. The computing unit 122 is configured to correct the motion blur of the current pair of raw skin and dermatoscopy images by (a) calculating the motion kernel from the relative motion vector and the estimated FOV 3D map, and (b) correcting the motion blur of the skin and dermatoscopy images using the current motion kernel.

[0151] At step 905, each captured image can be deblurred utilizing a respective section of the topography map using any suitable technique. For example, the computing unit 122 is configured to focus-deblur the skin image by (a) using first CNN-based 3D deblur algorithm and two FOV depth maps (estimated and estimated-modified) for generating two partially de-blurred skin images; and (b) using second CNN-based algorithm and the two partially deblurred images for generating a deblurred skin image and an improved FOV 3D depth map. In one or more embodiments, each captured image can be deblurred by preparing a first image through deblurring the captured image utilizing the improved 3D FOV depth map, modifying the 3D FOV map, and preparing a second image through deblurring the captured image utilizing the modified 3D FOV depth map. The first and second images can be blended into an enhanced resolution image. In one or more embodiments, the improved 3D FOV depth map is prepared using data that include overlapping sections of the previous 3D FOV depth maps.

[0152] The computing unit 122 can be configured to correct the 3D blur of the dermatoscopy image using the second CNN-based algorithm and the improved FOV 3D depth map calculated for the paired skin image.

[0153] At step 906, the computing unit 122 can be configured to enhance the current skin image resolution using a suitable MFSR algorithm and a sequence of at leastthree deblurred skin images (previous, current, and next). The computing unit 122 can further be configured to enhance the current dermatoscopy image resolution using the MFSR algorithm and a sequence of at least three deblurred dermatoscopy images (e.g., previous current and next).

[0154] At step 907, the computing unit 122 can be configured to prepare the estimated FOV depth map for deblurring the upcoming raw image pair, by (a) adding a 3D strip (e.g., about 0.5 mm width) perpendicular to the scan direction from the stored 3D surface map 412; (b) stitching the 3D strip with the previous improved 3D FOV depth map, as to approach continuity; c) adjusting the morphing kernel using the stitched 3D FOV depth map and the current image acquisition assembly 5DOF, and (d) morphing the stitched 3D image into the estimated FOV depth map using the adjusted kernel.

[0155] At each transition event (e.g., face to neck), the estimated 3D FOV depth map is calculated per step 904 instead of step 907. If the 3D FOV depth map includes a non-continuous region, the region’s geometry may be converged through an iterative process (e.g., by using several consecutive depth maps).

[0156] At step 908, the computing unit 122 can be configured to stitch the deblurred captured skin images into continuous high resolution skin images. For example, the computing unit 122 can be configured to stitch the deblurred skin image sequence into a large skin image, by: (a) calculating a reverse 3D kernel using the improved FOV 3D depth map; (b) back-morphing each high resolution skin image into its equivalent photogrametric high resolution skin image; and (c) stitching the sequence of photogrametric high resolution skin images into a continuous high resolution skin image, using any suitable CNN-based stitching algorithm.

[0157] In one or more embodiments, the computing unit 122 can be configured to tailor a deblurred dermatoscopy image sequence into a continuous dermatoscopy image by: (a) back-morphing each high resolution dermatoscopy image into its equivalent photogrametric high resolution dermatoscopy image using the said back morph kernel; and b) stitching the sequence of photogrametric high resolution dermatoscopy images into a continuous high resolution dermatoscopy image using said stitching algorithm with stitching instructions similar to those used for stitching the skin images.

[0158] In one or more embodiments, the computing unit 122 can be configured tostitch a sequence of improved FOV depth maps into a continuous depth map by (a) back- morphing each improved FOV 3D depth map into its equivalent photogrametric depth map, and b) stitching the sequence of morphed photogrametric depth maps into a continuous depth map using said stitching algorithm with stitching instructions similar to those used for stitching the skin images.

[0159] At step 909, the computing unit 122 is configured to process at least the continuous skin image processed at step 908 by executing a scanning algorithm that checks whether detected skin structures (e.g. actinic keratosis, moles, skin marks, etc.), within the continuous skin image meet a criterion or criteria.

[0160] If a suspicious skin structure is detected during the scan, its vicinity is extracted from the continuous depth map and examined for validating the finding. Each skin structure finding and its XYZ data can be recorded.

[0161] At step 910, the computing unit 122 can be configured to execute a lesion algorithm that scans at least the continuous dermatoscopy image for suspected subsurface skin structure. The algorithm is configured to classify each suspected sub-surface sub-skin structure as being one of a normal or an abnormal skin condition, and as a result, provides a value of potential malignancy of a given lesion. For example, a list of structures with potential malignancy are reviewed by Marghoob et. al., “Dermoscopy: A Review of the Structures That Facilitate Melanoma Detection,” J Am Osteopath Assoc, vol. 119 no. 6 , p. :380-390, 2019. Each suspected sub-surface skin structure and its XYZ coordinates are recorded, together with the respective map sections taken from the continuous skin image and the continuous depth map. The depth map section is also recorded, so that coincidence of sunken or raised regions with a sub-surface finding enhance the potential malignancy diagnosis.

[0162] In certain aspects, the algorithm can detect suspicious structures as follows: (a) detect all suspected skin structures in each captured image set; (b) convert each suspected skin structure into its embedded words representation; and (c) compare each embedded word with embedded words data base that use a collection of skin images and a list of structures, such as that of Magroob. See, e.g., X. Hu et. Al., “Enhancing Skin Disease Diagnosis: Interpretable Visual Concept Discovery with SAM Empowerment,” Arvix Sept. 14 (2024).

[0163] At step 911, the computing unit 122 can be configured to provide at least two lists of suspected findings, the first extracted at least from the continuous skin image and the second at least from the continuous dermatoscopy image. Each finding in the lists is accompanied by the relevant continuous image (and / or map) section. In this sense, a physician / doctor can later examine images / maps from each list e.g., utilizing a display, and make a diagnosis of that skin lesion.

[0164] With reference now to FIG. 11, a flowchart illustrates another example embodiment of processing steps executed by a computing unit. This method is initiated once a subject 103 to be examined is accommodated within the support frame 102 of the dermatoscopy system 100, and further once the image acquisition assembly 106 is ready to move towards a first body part (e.g. the torso) of a given body side (e.g. the ventral side of the body) of the subject 103.

[0165] At step 1001, the computing unit 122 activates a displacement element (e.g., actuators 108, 110) to move the 3D mapping assembly and acquire 3D mapping images of at least a portion of the body side of the subject. At step 1002, the computing unit 122 analyzes the acquired 3D images and prepares a scanning plan for a dermatoscopic camera of the first portion of exposed body side of the subject. The plan considerations include effort to preserve the working distance during the scan.

[0166] At step 1003, the computing unit 122 executes the scanning plan that includes a motion plan to automatically activate the displacement element (e.g., actuators 108, 110 and support 112). The scanning plan includes an imaging plan with instructions to the image acquisition assembly 106 acquires a set of raw dermatoscopy images while the camera 212 hovers at near-working distance, e.g., at approximately 50 - 150 mm. During the scan, the liquid lens 218 is adjusted to maintain focus whenever the planned working distance deviates from the nominal working distance. The image sensors 214 and 222 acquire sets of images while hovering at around the working distance.

[0167] At step 1004 the computing unit 122 receives data from the location sensor 115 and executes a 3D algorithm which calculates the FOV center 3D coordinates on the 3D map of the portion of the subject’s body and the estimated working distance. That data is recorded with each raw dermatoscopy image.

[0168] In certain embodiments of the present disclosure, the deviation and driftaccumulated during motion plan executed through the displacement element can be corrected, e.g., as shown in step 1005 in FIG. 11. This correction involves the computing unit 122 comparing the five DOF data provided by the location sensor 115 to the respective coordinate values calculated by the 3D scanning plan. If a drift is detected, the scan plan is corrected accordingly.

[0169] In other embodiments, the deviation is estimated by comparing the distance sensor readout to the location sensor 115 five DOF data. An above-threshold deviation of certain DOF is corrected for example by adding a constant increments to the relevant DOFs of the motion plan before the next image capture.

[0170] At step 1006, the computing unit processes the raw dermatoscopy images stream from the image sensor 214 (including the 3D coordinates) and uses an algorithm (e g., Multi-Frame Super Resolution or MSFR) for generating high resolution images thereof. Next, the computing unit processes the raw skin images stream from the image sensor 222 (including the 3D coordinates) and uses an algorithm (e.g., Multi-Frame Super Resolution or MSFR) for generating high resolution skin images thereof. Next, the computing unit execute a tailoring algorithm and uses visible images from image sensor 222 for tailoring the high resolution dermatoscopy images into a "continuous dermatoscopy image" of at least a portion of the user's skin.

[0171] At step 1007, the computing unit 122 processes the high resolution frames processed from raw visible images from image sensor 222 by executing a scanning algorithm that checks whether skin structures (e.g. actinic keratosis, moles, skin marks, etc.) included in said skin image(s) fulfill a given criterion. The findings for each skin image are recorded with that image.

[0172] At step 1008, the computing unit 122 executes a lesion algorithm which detects a suspected sub-surface skin structure and performs, at least, a classification task, and as a result, provides a value of “potential malignity” of a given lesion. The result is recorded in each of the high resolution dermatoscopy images.

[0173] At step 1009, the computing unit 122 provides two lists of images with suspected findings, one for the skin images and one for the high resolution dermatoscopy images. In this sense, a physician / doctor can later examine images from each list e.g., though a display, and make an accurate diagnosis of the skin lesion.

[0174] In the following, an example configuration of the medical skin imaging system 100 that may use hardware and methods is described. The system 100 in this example is designed for a scanning speed, e.g., of at least about 5 cm / sec and no greater than about 50 cm / sec. In one or more embodiments, the scanning speed of the system can be about 10 cm / sec on the subject skin. The FOV size on the skin at the nominal scan distance (lens front - skin distance ~70 mm) is 15* 20 mm (equivalent to 0.3 mA2 during scan period of 3 minutes). In certain embodiments, the NIR light source 204 illuminates the skin FOV with 1000 lumen / cmA2 during 0.5 msec (without polarizer) at telescope 216 setting of f / # = 5.6.

[0175] The attenuation provided by the cross polarizer 220 installed on the camera telescope 216 dramatically reduces the photon density incident on the image sensor 214. The above illumination parameters are, however, equivalent to those used by Fricke (2019), which demonstrated capture of contrast low-noise image. During the illumination assembly 200 illumination pulse, the FOV center moves ~50 microns (motion blur) on the subject’s skin surface 105 at the nominal scan distance.

[0176] Prior to the subject’s body scan, the displacement element (e.g., actuators 108, 110 and support 112) moves the 3D mapping assembly 114 to map at least a portion of the fixed subject’s body side facing the image acquisition assembly 106. At each 3D mapping event, the location sensor 115 readout (converted into 5DOF data) is recorded. Shortly after 3D mapping of at least a portion of the subject’s body, a scanning plan 412 is prepared. During the scan, the displacement element (e.g., actuators 108, 110 and support 112) moves the image acquisition assembly 106 along a scanning plan 412 trajectory so that the FOV center moves, e.g., at about 10 cm / sec on the subject’s skin. At a camera frame rate of 200 FPS, the FOV center points lie on a line with a distance of 0.5 mm between center points. In practice, the FOV center point will deviate from the scanning plan 412 locations. The short term pointing spread can include at least one of the following: displacement element speed error, displacement element angular error, or the natural motion of subject’s body.

[0177] At each image capture event, the image acquisition assembly 106 captures the skin and dermatoscopy images near-simultaneously. Each image pair is stored along with the illumination assembly 5DOF coordinates at the image capture event, calculatedfrom the location sensor 115 readout. In one or more embodiments, at least 10 percent of the image capture events from the location sensor readout is converted into image acquisition assembly 5DOF coordinates.

[0178] The captured image sequences (skin, dermatoscopy) may be converted into continuous high resolution images by (a) extracting a FOV-compatible depth map section from the stored 3D subject’s body map using the 5DOF data and the view pyramid; (b) utilizing the depth map and view pyramid for calculating: (i) the morph kernel; and (ii) the motion kernel, using the relative motion vector; (c) converting the FOV depth map into a 3D FOV depth map using the morph kernel; (d) correcting the motion blur of the raw skin and dermatoscopy image pair using the motion kernel; (e) sharpening the 3D blur of the corrected skin image through two cycles of two suitable CNN de-blur algorithms, and generating an improved FOV depth map; e) sharpening the dermatoscopy image using the said CNN algorithms and the improved FOV depth map; f) enhancing the sharpened skin and dermatoscopy image pair using an MF SR algorithm and an appropriate deblurred image sequence; (g) preparing the 3D FOV depth map for the upcoming image pair; (h) calculating the (photogrametric) morph kernel using the improved FOV depth map; (i) morphing each skin and dermatoscopy image pair and 3D FOV map into equivalent photogrametric images and map, using the photogrametric morph kernel; and (j) stitching sequences of morphed skin and dermatoscopy images and improved 3D FOV depth maps into a continuous skin and dermatoscopy images and continuous 3D depth map.

[0179] The location sensor 115 provides the 5DOF of the image acquisition assembly 106 with estimated angular accuracy of 500 microRadians. In turn, the estimated depth map edges may deviate up to 30 microns, which is negligible vs. the breathing cycle induced deviation. The CNN based de-blur processing is expected to correct the skin image to pixel-level resolution. For example, each pixel image covers a 10 micron skin rectangle when capturing the FOV on a 5 Mega pixels imager. Thus, it is expected that 3D deblurred images resolution would reach 15 microns, at the skin surface.

[0180] The MFSR process is expected to remove residual blur (e.g., motion and 3D blur) to the pixel level. An example for MFSR algorithm and results is described byZ. Ma (2014). Due to the fast capture rate (200 Hz), about 40 frames share certain fraction in the center (e.g., 20th) frame FOV. Under such conditions, the MFSR processed discrete skin image resolution is expected to reach 10 microns. In turn, fine details in Dermatoscopy images (if exist) are also expected to be restored.

[0181] Embodiments of the disclosure are defined in the claims; however, herein here is provided a non-exhaustive listing of non-limiting Clauses. Any one or more of the features of these Clauses may be combined with any one or more features of another Clause, embodiment, or aspect described herein.

[0182] Clause 1. An imaging system comprising: an image acquisition assembly comprising a camera and imaging optics configured to capture photographic images of a skin surface of a mammal subject disposed on a support structure while executing a scanning plan; an actuator coupled to the image acquisition assembly, the actuator configured to move the image acquisition assembly over the subject at a near-nominal working distance; a location sensor configured to detect location and orientation of the image acquisition assembly while capturing overlapping photographic images, wherein the location sensor comprises at least one light source configured to direct an electromagnetic beam to a target surface; a three-dimensional (3D) mapping assembly that obtains a 3D topography map representative of a 3D surface geometry of the subject; an illumination assembly configured to illuminate a camera field of view so that the captured photographic images comprise medical dermatoscopic information; and a computing unit coupled to the image acquisition assembly, the actuator, and the 3D mapping assembly, the computing unit configured to: develop a scanning plan based on the topography map, the scanning plan comprising a traversal path of the image acquisition assembly and a camera triggering order along the traversal path, the traversal path maintaining the near-nominal working distance between the skin surface and the image acquisition assembly and a view axis of the image acquisition assembly is nearnormal to the viewed captured skin surface; execute the scanning plan on-the-fly via the actuator and trigger the image acquisition assembly to obtain at least one sequence of significantly overlapping images along the traversal path; and deblur at least one sequence of significantly overlapping images utilizing the 3D topography map.

[0183] Clause 2. The imaging system of Clause 1, wherein the location sensorcomprises a LiDAR device fixed to the image acquisition assembly, wherein the LiDAR device is configured to direct an electromagnetic energy beam along a LiDAR view axis to a target surface fixed to the support structure of the system and comprising a pattern, wherein the LiDAR view axis defines an angle with the view axis of the image acquisition assembly of at least about 30 degrees and no greater than about 90 degrees.

[0184] Clause 3. The imaging system of Clause 2, wherein the target surface is substantially perpendicular to the mammal subject's spinal column.

[0185] Clause 4. The imaging system of any one of Clauses 2-3 wherein the target surface is disposed adjacent to the subject.

[0186] Clause 5. The imaging system of Clause 1, wherein the location sensor comprises: a screen fixed to a support structure; an emitter fixed to the image acquisition assembly and configured to illuminate the screen with a structured image; and a camera fixed to the support structure and configured to capture at least a portion of the screen while the screen is illuminated by the emitter.

[0187] Clause 6. The imaging system of any one of Clauses 1-5, wherein the computing unit is further configured to convert a location sensor readout from the location sensor into image acquisition assembly five degree of freedom (5DOF) coordinates for each captured photographic image.

[0188] Clause 7. The imaging system of Clause 6, wherein the computing unit is further configured to determine an initial 3D field of view depth map utilizing the image acquisition assembly 5DOF coordinates and the topography map.

[0189] Clause 8. The imaging system of any one of Clauses 1-7, wherein the illumination assembly comprises a polarizing filter and wherein the image acquisition assembly further comprises a polarization filter having a 90 degree offset relative to the illumination assembly polarizing filter.

[0190] Clause 9. The imaging system of any one of Clauses 6-8, wherein the computing unit is further configured to utilize at least two overlapping sequential images of the captured photographic images, their respective image acquisition assembly 5DOF coordinates and the 3D field of view depth map to provide at least one high resolution image.

[0191] Clause 10. The imaging system of any one of Clauses 6-9, wherein theimage acquisition assembly further comprises a focus adjustment unit connected to the imaging optics of the image acquisition assembly, wherein the computing unit is further configured to adjust imaging optics focus based upon the 5D0F coordinates and the scanning plan.

[0192] Clause 11. The imaging system of any one of Clauses 1-10, wherein the imaging optics comprise a metalens.

[0193] Clause 12. The imaging system of any one of Clauses 1-11, wherein the image acquisition assembly is configured to capture the photographic images of the subject surface on-the-fly at a high image capture events rate of at least about 15 Hz and no greater than about 600 Hz.

[0194] Clause 13. The imaging system of any one of Clauses 1-12, wherein the image acquisition assembly is configured to capture the photographic images at a short working distance from the subject of at least about but not more than about .

[0195] Clause 14. The imaging system of any one of Clauses 1-13, wherein a scanning speed of the image acquisition system is at least about 5 cm / sec and no greater than about 50 cm / sec.

[0196] Clause 15. The imaging system of any of Clauses 1-14, wherein the image acquisition assembly further comprises: a dermatoscopic imaging path comprising a dermatoscopic imaging sensor; and a skin imaging path comprising a skin imaging sensor; wherein the imaging optics comprises a telescope that is common for the dermatoscopic imaging path and the skin imaging path and wherein the significantly overlapping images obtained by the image acquisition assembly comprise dermatoscopy and skin images captured near-simultaneously by the dermatoscopic imaging sensor and the skin imaging sensor.

[0197] Clause 16. The imaging system of Clause 15, wherein the computing unit is further configured to update a 3D FOV depth map using the captured skin image and deblur the captured dermatoscopy image utilizing the updated 3D FOV map.

[0198] Clause 17. The imaging system of any one of Clauses 15-16, wherein the computing unit is further configured to execute a lesion algorithm on the captured dermatoscopy images that classifies each dermatoscopy image as being one of a normal or an abnormal skin condition.

[0199] Clause 18. The imaging system of any one of Clauses 15-17, wherein the computing unit is further configured to execute a skin structure algorithm that scans the captured skin images to determine whether any one of a plurality of detected skin structures meet a criteria.

[0200] Clause 19. The imaging system of Clause 18, wherein the computing unit is further configured to execute at least the skin structure algorithm on images from a region of the subject surface of the mammal subject, wherein the images are selected from a stitched skin image, a stitched dermatoscopy image, or a stitched depth map.

[0201] Clause 20. The imaging system of any one of Clauses 1-19, further comprising a privacy cover that covers the subject and image acquisition assembly, the privacy cover configured to minimize stray light from entering into skin of the subj ect and the imaging optics of the image acquisition assembly during execution of the scanning plan.

[0202] Clause 21. The imaging system of any one of Clauses 1-20, wherein the illumination assembly is further configured to enhance informative structures beneath skin of the subject within the captured images.

[0203] Clause 22. The imaging system of any one of Clauses 1-21, wherein the actuator is further configured to manipulate the image acquisition assembly within five degrees of freedom, and wherein the scanning plan includes five degree of freedom instructions according to the 3D topography map.

[0204] Clause 23. A method comprising: positioning a mammal subject on a support structure of an imaging system so that a skin surface of the mammal subject is visible by an image acquisition assembly of the imaging system comprising a camera configured to capture photographic images of the mammal subject on-the-fly; obtaining a topography map representative of a three-dimensional (3D) surface geometry of the mammal subject using a 3D mapping assembly of the imaging system; developing a scanning plan based on the topography map, the scanning plan comprising a traversal path of the image acquisition assembly and an image capture event triggering order along the traversal path, the traversal path maintaining a near-nominal working distance between the skin surface and the camera and further maintaining optical axis orientation of the camera near normal to skin FOV of the surface of the mammal subject during eachimage capture; executing the scanning plan via an actuator, a location sensor, and the image acquisition assembly; simultaneously illuminating the subject and capturing significantly overlapping photographic images of the subject at each image capture event while executing the scanning plan, wherein the captured photographic images comprise captured dermatoscopy images and captured skin images; converting a location sensor readout from the location sensor into image acquisition assembly five degree of freedom (5D0F) coordinates for each captured photographic image; deblurring each captured image utilizing a respective section of the topography map; and stitching the deblurred captured photographic images into a continuous high-resolution dermatoscopy images and continuous high-resolution skin images.

[0205] Clause 24. The method of Clause 23, wherein at least 10 percent of the image capture events the location sensor readout is converted into image acquisition assembly 5DOF coordinates.

[0206] Clause 25. The method of any one of Clauses 23-24, wherein at each image capture event, the dermatoscopy and skin images are captured near- simultaneously as separate images, and wherein the illumination assembly is configured to emit a selected spectrum and intensity for the dermatoscopy image and for the skin image.

[0207] Clause 26. The method of any one of Clauses 23-25, wherein deblurring each captured photographic image comprises motion deblurring each captured image utilizing a relative motion vector calculated from a latest sequence of 5DOF coordinates and dedicated motion compensation software.

[0208] Clause 27. The method of any one of Clauses 23-26, wherein at least the skin images are captured utilizing a metalens.

[0209] Clause 28. The method of any one of Clauses 23-27, further comprising preparing an improved 3D FOV depth map utilizing data selected from the topography map, previous depth maps, and a depth map generated utilizing the captured skin image, and current 5DOF coordinates.

[0210] Clause 29. The method of Clause 28, wherein deblurring each captured image comprises: preparing a first image through deblurring the captured image utilizing the improved 3D FOV depth map; modifying the 3D FOV depth map; and preparing asecond image through deblurring the captured image utilizing the modified 3D FOV depth map; wherein the first and second images are blended into an enhanced resolution image.

[0211] Clause 30. The method of any one of Clauses 28-29, wherein the improved 3D FOV depth map is prepared using data that comprises overlapping sections of the previous 3D FOV depth maps.

[0212] Clause 31. The method of Clause 30, wherein deblurring each captured image further comprises deblurring each captured photographic image utilizing the improved 3D depth maps.

[0213] Clause 32. The method of any one of Clauses 23-31, wherein each captured photographic image is deblurred utilizing a CNN algorithm.

[0214] Clause 33. The method any one of Clauses 23-32, wherein at least one deblurred skin image sequence is stitched into a continuous skin image utilizing a stitching algorithm.

[0215] Clause 34. The method of any one of Clauses 23-33, wherein at least one deblurred dermatoscopy image sequence is stitched into a continuous dermatoscopy image utilizing a stitching algorithm.

[0216] Clause 35. The method of any one of Clauses 23-34, wherein motion deblurring, focus deblurring, and stitching continuous images utilize a photogrametric grid.

[0217] Clause 36. An imaging system comprising: an image acquisition assembly comprising a camera and imaging optics operable to acquire photographic images of a subject while executing a scan plan; an actuator coupled to the image acquisition assembly, the actuator operable to move the camera over the subject at a nearpredetermined, nominal working distance; a location sensor operable to detect location and orientation of the camera while capturing overlapping images; a three-dimensional (3D) mapping assembly that obtains a topography map representative of a 3D surface geometry of the subject; an illumination system operable to illuminate a camera field of view so that the captured photographic images comprise medical dermatoscopic information; and a computing unit coupled to the image acquisition assembly, the actuator, and the 3D mapping assembly, the computing unit operable to: develop ascanning plan based on the topography map, the scanning plan comprising a traversal path of the image acquisition assembly and a camera triggering order along the traversal path, the traversal path maintaining near-nominal working distance between the subject and the imaging system so that an axis of the imaging system is near-normal to the captured subject surface; execute the scanning plan via the actuator and triggering the image acquisition assembly to obtain overlapping images along the traversal path; and combine the overlapping images taken by the image acquisition assembly into one or more high-resolution dermatoscopy images.

[0218] Clause 37. The imaging system of Clause 36, wherein the location sensor comprises a distance sensor which provides a measured working distance while executing the scan plan.

[0219] Clause 38. The imaging system of Clause 37, wherein the computing unit is further operable to: compare the camera locations with orientation obtained by the location sensor versus corresponding locations and orientation provided by the scanning plan; and correct the scanning plan while it is being executed.

[0220] Clause 39. The imaging system of Clause 36, wherein the computing unit is further operable to: register the camera location and orientation data for each of the overlapping images in respect to the topography map; and use at least two sequential overlapping images with their respective registration data into a high-resolution dermatoscopy image.

[0221] Clause 40. The imaging system of Clause 36, wherein the illumination system is operable to enhance informative structures beneath skin of the subject within the captured images.

[0222] Clause 41. The imaging system of Clause 36, wherein the camera comprises two separate imaging sensors and separate optical paths, and wherein one of the separate optical paths excludes an optical filter.

[0223] Clause 42. The imaging system of Clause 41, wherein combining the overlapping images taken by the camera comprises using a Multi-Frame Super Resolution (MFSR) algorithm.

[0224] Clause 43. The imaging system of Clause 36, wherein the actuator provides five degree-of-freedom positioning of the camera.

[0225] Clause 44. The imaging system of Clause 36, wherein the imaging optics comprises a liquid lens, operable to conduct camera focus adjustments during executing the scan plan.

[0226] Clause 45. The imaging system of Clause 36, wherein the illumination system comprises a polarization filter having a 90° offset with respect to a polarization filter of the imaging optics.

[0227] Clause 46. The imaging system of Clause 36, wherein the nominal working distance is between and , inclusive.

[0228] Clause 47. The imaging system of Clause 36, wherein the computing unit is further operable to execute a lesion algorithm on the one or more high-resolution dermatoscopy images which classifies each high-resolution dermatoscopy image captured as being one of normal and abnormal skin condition.

[0229] Clause 48. The imaging system of Clause 36, wherein the computing unit is further operable to execute a skin structure algorithm which scans high resolution skin image to check whether any one of a plurality of detected skin structures fulfills a criterion.

[0230] Clause 49. The imaging system of Clause 36, further comprising a privacy cover that covers the subject and image scan system, the privacy cover operable to minimize stray light from entering into skin of the subject and the imaging optics during the execution of the scanning plan.

[0231] Clause 50. A method, comprising: positioning a subject on a support structure so that a surface of the subject is visible by an image acquisition assembly comprising a camera operable to acquire photographic images of the subject; obtaining a topography map representative of a three-dimensional (3D) surface geometry of the subject using a 3D mapping assembly; developing a scanning plan based on the topography map, the scanning plan comprising a traversal path of the image acquisition assembly and a camera triggering order along the traversal path, the traversal path maintaining a nominal working distance between the subject and the camera and further maintaining optical axis orientation near normal to skin field of view during each image capture; executing the scanning plan via an actuator and the image acquisition assembly to obtain overlapping images along the traversal path; and combining the overlappingimages taken by the image acquisition assembly into a sequence of high-resolution dermatoscopy and skin images.

[0232] Clause 51. The method of Clause 50, further comprising using a location sensor for determining location and orientation of the camera while taking each of the respective overlapping images.

[0233] Clause 52. The method of Clause 51, further comprising: comparing a readout of the location sensor with corresponding five degree of freedom data of the scanning plan; and correcting the scanning plan while it is being executed.

[0234] Clause 53. The method of Clause 51, further comprising: comparing a location sensor readout with corresponding distance in the scanning plan; and further correcting the scanning plan while it is being executed.

[0235] Clause 54. The method of Clause 51, further comprising: converting the locations and orientation from the location sensor, in respect to a corrected scan plan, into registration data for each of the overlapping images in respect to the topography map; and converting at least two sequential captured images with their respective registration data into a high-resolution dermatoscopy and skin image.

[0236] Clause 55. The method of Clause 50, wherein the camera and an illumination system are operable to enhance dermatoscopic information beneath skin of the subject as captured in the dermatoscopy images.

[0237] Clause 56. The method of Clause 50, wherein combining the overlapping images taken by the camera comprises using a Multi-Frame Super Resolution (MFSR) algorithm.

[0238] Clause 57. The method of Clause 50, wherein executing the scanning plan comprises providing five degree-of-freedom positioning of the camera.

[0239] Clause 58. The method of Clause 50, wherein the imaging system comprises a liquid lens, and wherein the scanning plan provides focus adjustments to the liquid lens along the traversal path.

[0240] Clause 59. The method of Clause 50, further comprising executing a lesion algorithm one or more high-resolution dermatoscopy images which classifies each high-resolution dermatoscopy image captured from as being one of normal and abnormal skin conditions.

[0241] Clause 60. The method of Clause 50, further comprising executing a skin structure algorithm which scans high resolution skin images to check whether any one of a plurality of detected skin structures fulfills a criterion.

[0242] Clause 61. The method of Clause 53, further comprising: obtaining, via the 3D mapping assembly, at least three topography maps representative of a 3D map of the subject while breathing; using the location sensor readout to predict a current phase of a breathing cycle of the subject; correct the traversal path based on the predicted breathing cycle to maintain near-nominal working distance between the subject and the camera during a next image capture; and based on the corrected traversal path and the predicted breathing cycle, provide a focus adjustment signal to a liquid lens to maintain focusing during the next image capture.

[0243] Clause 62. An imaging system comprising: an image acquisition assembly comprising a camera and imaging optics operable to capture photographic images of a subject surface of a subject while executing a scanning plan; an actuator coupled to the image acquisition assembly, the actuator operable to move the camera over the subject at a near-nominal working distance; a location sensor operable to detect location and orientation of the image acquisition assembly while capturing overlapping images; a three-dimensional (3D) mapping assembly that obtains a topography map representative of a 3D surface geometry of the subject; an illumination assembly operable to illuminate a camera field of view so that the captured photographic images comprise medical dermatoscopic information; and a computing unit coupled to the image acquisition assembly, the actuator, and the 3D mapping assembly, the computing unit operable to: develop a scanning plan based on the topography map, the scanning plan comprising a traversal path of the image acquisition assembly and a camera triggering order along the traversal path, the traversal path maintaining the near-nominal working distance between the subject and the image acquisition system so that an axis of the image acquisition assembly is near-normal to the captured subject surface; execute the scanning plan on- the-fly via the actuator and trigger the image acquisition assembly to obtain at least one sequence of significantly overlapping images along the traversal path; and deblur at least one sequence of significantly overlapping images utilizing the 3D topography map.

[0244] Clause 63. The imaging system of Clause 62, wherein the imageacquisition assembly is operable to capture photographic images of the subject surface on-the-fly at a high image capture events rate of at least about 15 Hz and no greater than about 600 Hz.

[0245] Clause 64. The imaging system of any one of Clauses 62-63, wherein the image acquisition assembly is operable to capture photographic images at a short working distance from the subject of less than about .

[0246] Clause 65. The imaging system of Clause 62, wherein the image acquisition assembly further comprises: a dermatoscopic imaging path comprising a dermatoscopic imaging sensor; a skin imaging path comprising a skin imaging sensor; and a telescope utilized in the dermatoscopy imaging path and the skin imaging path; wherein the image acquisition assembly is operable to near-simultaneously capture dermatoscopy and skin images using the dermatoscopic imaging sensor and the skin imaging sensor.

[0247] Clause 66. The imaging system of Clause 65, wherein the skin imaging path comprises a metalens.

[0248] Clause 67. The imaging system of any one of Clauses 62-66, wherein the illumination assembly comprises a polarizing filter and wherein the image acquisition assembly comprises a polarization filter having a 90 degree offset relative to the illumination assembly polarizing filter.

[0249] Clause 68. The imaging system of any one of Clauses 62-67, wherein the illumination assembly is further operable to enhance informative structures beneath the skin of the subject within the captured images.

[0250] Clause 69. The imaging system of any one of Clauses 62-68, wherein the location sensor comprises: a light projector fixed to the image acquisition assembly; a screen fixed to the support structure; and a camera, fixed to the support structure; wherein the camera is triggered at a significant fraction of the image capture events.

[0251] Clause 70. The imaging system of any of Clauses 62-69, wherein an illumination axis of light from the illumination assembly is substantially coaxial with the optical axis of the image acquisition assembly telescope.

[0252] Clause 71. The imaging system of any one of Clauses 62-70, wherein the nominal working distance is at least about and no greater than about .

[0253] Clause 72. The imaging system of any of Clauses 62-71, wherein the actuator is operable to manipulate the image acquisition assembly within five degrees of freedom, and wherein the scanning plan includes five degree of freedom instructions according to the stored topography map.

[0254] Clause 73. The imaging system of Clause 72, wherein the computing unit is operable to convert a location sensor readout into image acquisition assembly five degree of freedom (5DOF) coordinates.

[0255] Clause 74. The imaging system of Clause 73, wherein the computing unit is operable to determine an initial 3D field of view depth map utilizing the image acquisition assembly 5DOF coordinates and the stored topography map.

[0256] Clause 75. The imaging system of Clause 74, wherein the computing unit is further operable to utilize at least two sequential images with their respective image acquisition assembly 5DOF coordinates to provide at least one high resolution image.

[0257] Clause 76. The imaging system of any of Clauses 73-75, wherein the image acquisition assembly further comprises a focus adjustment unit connected to the imaging optics of the image acquisition assembly, wherein the computing unit is further operable to direct focus adjustment unit based upon the 5DOF coordinates and the scanning plan.

[0258] Clause 77. The imaging system of any one of Clauses 62-76, further comprising a privacy cover that covers the subject and image acquisition assembly, the privacy cover operable to minimize stray light from entering into skin of the subject and the imaging optics of the image acquisition assembly during the acquisition of the scanning plan.

[0259] Clause 78. The imaging system of any of Clauses 62-77, wherein a scanning speed of the image acquisition system is at least about 5 cm / sec and no greater than about 50 cm / sec.

[0260] Clause 79. The imaging system of any one of Clauses 62-78 wherein the computing unit is further operable to execute a lesion algorithm on the one or more high- resolution dermatoscopy images that classifies each high-resolution captured dermatoscopy image as being one of a normal or an abnormal skin condition.

[0261] Clause 80. The imaging system of any of Clauses 62-79, wherein thecomputing unit is further operable to execute a skin structure algorithm that scans high resolution skin images to check whether any one of plurality of detected skin structures fulfdls a criterion.

[0262] Clause 81. The imaging system of Clause 80, wherein the computing unit is further operable to execute at least the skin structure algorithm on a region, wherein the region data is taken from a stitched image selected from a skin image, a dermatoscopy image, or a depth map.

[0263] Clause 82. The imaging system of any one of Clauses 62-81, wherein the computing unit is further operable to update a 3D FOV depth map using the captured skin image and deblur the captured dermatoscopy image utilizing the updated 3D FOV map.

[0264] Clause 83. A method comprising: positioning a subject on a support structure of an imaging system so that a surface of the subject is visible by an image acquisition assembly of the imaging system comprising a camera operable to acquire photographic images of the subject; obtaining a topography map representative of a three-dimensional (3D) surface geometry of the subject using a 3D mapping assembly of the imaging system; developing a scanning plan based on the topography map, the scanning plan comprising a traversal path of the image acquisition assembly and a camera triggering order along the traversal path, the traversal path maintaining a nominal working distance between the subject and the camera and further maintaining optical axis orientation near normal to skin FOV during each image capture; executing the scanning plan via an actuator of the imaging system, the location sensor, and the image acquisition assembly to obtain significantly overlapping images along the traversal path; deblurring each captured image utilizing the respective section of the topography map; and combining the overlapping images taken by the image acquisition assembly into a sequence of high-resolution dermatoscopy and skin images.

[0265] Clause 84. The method of Clause 83, wherein at a significant fraction of the image capture events and the location sensor readout are converted into image acquisition assembly 5DOF coordinates.

[0266] Clause 85. The method of any of Clauses 83-84, wherein at each image capture event, skin and dermatoscopy images of a near similar FOV are captured near- simultaneously on separate images, and wherein the illumination assembly is configuredto emit a selected spectrum and intensity for each image type.

[0267] Clause 86. The method of any of Clauses 83-85, wherein deblurring each captured image comprises motion deblurring each captured image utilizing a dedicated motion compensation software and the relative motion vector data.

[0268] Clause 87. The method of any of Clauses 83-86, further comprising preparing an improved 3D FOV depth map utilizing the data selected from the stored 3D subject's body topography, previous depth maps, a depth map generated along the deblurring of the captured skin image, and current 5DOF data.

[0269] Clause 88. The imaging method of Clause 87, further comprising: preparing a first deblurred images utilizing the improved FOV depth map; and preparing a second deblurred image utilizing a modified 3D FOV depth map; wherein the first and second images are blended into an enhanced resolution image using the deblurring software.

[0270] Clause 89. The method of any one of Clauses 83-87, wherein a skin image is captured by the image acquisition assembly comprising a metalens with high chromatic dispersion and wherein at least the capturing imager comprises an array of chromatic filters.

[0271] Clause 90. The method of any one of Clauses 83-89, wherein an improved 3D FOV depth map is prepared using data that comprises overlapping sections of the previous depth maps.

[0272] Clause 91. The method of Clause 90, wherein deblurring each captured image utilizes co-generated improved 3D depth maps for deblurring the co-captured image.

[0273] Clause 92. The method of any one of Clauses 83-91, further comprising preparing an improved 3D FOV depth map using data comprising overlapping sections of previous depth maps.

[0274] Clause 93. The method of any of Clauses 83-92, wherein deblurring each captured image utilizes a CNN algorithm.

[0275] Clause 94. The method any one of Clauses 83-93, wherein at least one skin image sequence is tailored into a large image using a stitching algorithm.

[0276] Clause 95. The method of any one of Clauses 83-94, wherein at least onedermatoscopy image sequence is tailored into a large image using a stitching algorithm.

[0277] Clause 96. The method of any of Clauses 83-95, wherein the motion- deblur, focus deblur, and image stitching processes are dominantly conducted using a photogrametric grid.

[0278] The various embodiments described above may be implemented using circuitry, firmware, and / or software modules that interact to provide particular results. One of skill in the arts can readily implement such described functionality, either at a modular level or as a whole, using knowledge generally known in the art. For example, the flowcharts and control diagrams illustrated herein may be used to create computer- readable instructions / code for execution by a hardware processor. Such instructions may be stored on a non-transitory computer-readable medium and transferred to the processor for execution as is known in the art. The structures and procedures shown above are only a representative example of embodiments that can be used to provide the functions described hereinabove.

[0279] Unless otherwise indicated, all numbers expressing feature sizes, amounts, and physical properties used in the specification are to be understood as being modified in all instances by the term “about.” Accordingly, unless indicated to the contrary, the numerical parameters set forth in the foregoing specification are approximations that can vary depending upon the desired properties sought to be obtained by those skilled in the art utilizing the teachings disclosed herein. The use of numerical ranges by endpoints includes all numbers within that range (e.g. 1 to 5 includes 1, 1.5, 2, 2.75, 3, 3.80, 4, and 5) and any range within that range.

[0280] The terms “coupled” or “connected” refer to elements being attached to each other either directly (in direct contact with each other) or indirectly (having one or more elements between and attaching the two elements). Either term may be modified by “operatively” and “operably,” which may be used interchangeably, to describe that the coupling or connection is configured to allow the components to interact to carry out at least some functionality.

[0281] Terms related to orientation, such as “top,” “bottom,” “side,” and “end,” are used to describe relative positions of components (e.g., as arranged in the figures) and are not meant to limit the orientation of the embodiments contemplated. For example, anembodiment described as having a “top” and “bottom” also encompasses embodiments thereof rotated in various directions unless the content clearly dictates otherwise.

[0282] Reference to “one embodiment,” “an embodiment,” “certain embodiments,” or “some embodiments,” etc., means that a particular feature, configuration, composition, or characteristic described in connection with the embodiment is included in at least one embodiment of the disclosure. Thus, the appearances of such phrases in various places throughout are not necessarily referring to the same embodiment of the disclosure. Furthermore, the particular features, configurations, compositions, or characteristics may be combined in any suitable manner in one or more embodiment.

[0283] References to a “combination” of different elements is also meant to include each element on its own unless otherwise indicated. For example, a combination of A, B, and C may include any one of A, B, or C alone, as well as A+B, A+C, A+B+C, etc. Further, where the elements of the combinations are actions (e.g., steps of a method), the listing of actions is not meant to imply a specific order that the actions may be taken in the combination unless otherwise indicated.

Claims

What is claimed is:

1. An imaging system comprising: an image acquisition assembly comprising a camera and imaging optics configured to capture photographic images of a skin surface of a mammal subject disposed on a support structure while executing a scanning plan; an actuator coupled to the image acquisition assembly, the actuator configured to move the image acquisition assembly over the subject at a near-nominal working distance; a location sensor configured to detect location and orientation of the image acquisition assembly while capturing overlapping photographic images, wherein the location sensor comprises at least one light source configured to direct an electromagnetic beam to a target surface; a three-dimensional (3D) mapping assembly that obtains a 3D topography map representative of a 3D surface geometry of the subject; an illumination assembly configured to illuminate a camera field of view so that the captured photographic images comprise medical dermatoscopic information; and a computing unit coupled to the image acquisition assembly, the actuator, and the 3D mapping assembly, the computing unit configured to: develop a scanning plan based on the topography map, the scanning plan comprising a traversal path of the image acquisition assembly and a camera triggering order along the traversal path, the traversal path maintaining the near- nominal working distance between the skin surface and the image acquisition assembly and a view axis of the image acquisition assembly is near-normal to the viewed captured skin surface; execute the scanning plan on-the-fly via the actuator and trigger the image acquisition assembly to obtain at least one sequence of significantly overlapping images along the traversal path; and deblur at least one sequence of significantly overlapping images utilizing the 3D topography map.

2. The imaging system of claim 1, wherein the location sensor comprises a LiDAR device fixed to the image acquisition assembly, wherein the LiDAR device is configured to direct an electromagnetic energy beam along a LiDAR view axis to a target surface fixed to the support structure of the system and comprising a pattern, wherein the LiDAR view axis defines an angle with the view axis of the image acquisition assembly of at least about 30 degrees and no greater than about 90 degrees.

3. The imaging system of claim 2, wherein the target surface is substantially perpendicular to the mammal subject’s spinal column.

4. The imaging system of claim 1, wherein the location sensor comprises: a screen fixed to a support structure; an emitter fixed to the image acquisition assembly and configured to illuminate the screen with a structured image; and a camera fixed to the support structure and configured to capture at least a portion of the screen while the screen is illuminated by the emitter.

5. A method comprising: positioning a mammal subject on a support structure of an imaging system so that a skin surface of the mammal subject is visible by an image acquisition assembly of the imaging system comprising a camera configured to capture photographic images of the mammal subject on-the-fly; obtaining a topography map representative of a three-dimensional (3D) surface geometry of the mammal subject using a 3D mapping assembly of the imaging system; developing a scanning plan based on the topography map, the scanning plan comprising a traversal path of the image acquisition assembly and an image capture event triggering order along the traversal path, the traversal path maintaining a near-nominal working distance between the skin surface and the camera and further maintaining optical axis orientation of the camera near normal to skin FOV of the surface of the mammal subject during each image capture;executing the scanning plan via an actuator, a location sensor, and the image acquisition assembly; simultaneously illuminating the subject and capturing significantly overlapping photographic images of the subject at each image capture event while executing the scanning plan, wherein the captured photographic images comprise captured dermatoscopy images and captured skin images; converting a location sensor readout from the location sensor into image acquisition assembly five degree of freedom (5D0F) coordinates for each captured photographic image; deblurring each captured image utilizing a respective section of the topography map; and stitching the deblurred captured photographic images into a continuous high- resolution dermatoscopy images and continuous high-resolution skin images.

6. The method of claim 5, wherein at each image capture event, the dermatoscopy and skin images are captured near-simultaneously as separate images, and wherein the illumination assembly is configured to emit a selected spectrum and intensity for the dermatoscopy image and for the skin image.

7. An imaging system comprising: an image acquisition assembly comprising a camera and imaging optics operable to acquire photographic images of a subject while executing a scan plan; an actuator coupled to the image acquisition assembly, the actuator operable to move the camera over the subject at a near-predetermined, nominal working distance; a location sensor operable to detect location and orientation of the camera while capturing overlapping images; a three-dimensional (3D) mapping assembly that obtains a topography map representative of a 3D surface geometry of the subject; an illumination system operable to illuminate a camera field of view so that the captured photographic images comprise medical dermatoscopic information; anda computing unit coupled to the image acquisition assembly, the actuator, and the 3D mapping assembly, the computing unit operable to: develop a scanning plan based on the topography map, the scanning plan comprising a traversal path of the image acquisition assembly and a camera triggering order along the traversal path, the traversal path maintaining near- nominal working distance between the subject and the imaging system so that an axis of the imaging system is near-normal to the captured subject surface; execute the scanning plan via the actuator and triggering the image acquisition assembly to obtain overlapping images along the traversal path; and combine the overlapping images taken by the image acquisition assembly into one or more high-resolution dermatoscopy images.

8. The imaging system of claim 7, wherein the location sensor comprises a distance sensor which provides a measured working distance while executing the scan plan.

9. The imaging system of claim 8, wherein the computing unit is further operable to: compare the camera locations with orientation obtained by the location sensor versus corresponding locations and orientation provided by the scanning plan; and correct the scanning plan while it is being executed.

10. The imaging system of claim 7, wherein the computing unit is further operable to: register the camera location and orientation data for each of the overlapping images in respect to the topography map; and use at least two sequential overlapping images with their respective registration data into a high-resolution dermatoscopy image.

11. A method, comprising: positioning a subject on a support structure so that a surface of the subject is visible by an image acquisition assembly comprising a camera operable to acquire photographic images of the subject;obtaining a topography map representative of a three-dimensional (3D) surface geometry of the subject using a 3D mapping assembly; developing a scanning plan based on the topography map, the scanning plan comprising a traversal path of the image acquisition assembly and a camera triggering order along the traversal path, the traversal path maintaining a nominal working distance between the subject and the camera and further maintaining optical axis orientation near normal to skin field of view during each image capture; executing the scanning plan via an actuator and the image acquisition assembly to obtain overlapping images along the traversal path; and combining the overlapping images taken by the image acquisition assembly into a sequence of high-resolution dermatoscopy and skin images.

12. The method of claim 11, further comprising using a location sensor for determining location and orientation of the camera while taking each of the respective overlapping images.

13. An imaging system comprising: an image acquisition assembly comprising a camera and imaging optics operable to capture photographic images of a subject surface of a subject while executing a scanning plan; an actuator coupled to the image acquisition assembly, the actuator operable to move the camera over the subject at a near-nominal working distance; a location sensor operable to detect location and orientation of the image acquisition assembly while capturing overlapping images; a three-dimensional (3D) mapping assembly that obtains a topography map representative of a 3D surface geometry of the subject; an illumination assembly operable to illuminate a camera field of view so that the captured photographic images comprise medical dermatoscopic information; and a computing unit coupled to the image acquisition assembly, the actuator, and the 3D mapping assembly, the computing unit operable to:develop a scanning plan based on the topography map, the scanning plan comprising a traversal path of the image acquisition assembly and a camera triggering order along the traversal path, the traversal path maintaining the near- nominal working distance between the subject and the image acquisition system so that an axis of the image acquisition assembly is near-normal to the captured subject surface; execute the scanning plan on-the-fly via the actuator and trigger the image acquisition assembly to obtain at least one sequence of significantly overlapping images along the traversal path; and deblur at least one sequence of significantly overlapping images utilizing the 3D topography map.

14. The imaging system of claim 13, wherein the image acquisition assembly further comprises: a dermatoscopic imaging path comprising a dermatoscopic imaging sensor; a skin imaging path comprising a skin imaging sensor; and a telescope utilized in the dermatoscopy imaging path and the skin imaging path; wherein the image acquisition assembly is operable to near-simultaneously capture dermatoscopy and skin images using the dermatoscopic imaging sensor and the skin imaging sensor.

15. The imaging system of claim 14, wherein the skin imaging path comprises a metalens.

16. The imaging system of any of claims 13—15, wherein the actuator is operable to manipulate the image acquisition assembly within five degrees of freedom, and wherein the scanning plan includes five degree of freedom instructions according to the stored topography map.

17. The imaging system of claim 16, wherein the computing unit is operable to convert a location sensor readout into image acquisition assembly five degree of freedom (5D0F) coordinates.

18. The imaging system of claim 17, wherein the computing unit is operable to determine an initial 3D field of view depth map utilizing the image acquisition assembly 5D0F coordinates and the stored topography map.

19. A method comprising: positioning a subject on a support structure of an imaging system so that a surface of the subject is visible by an image acquisition assembly of the imaging system comprising a camera operable to acquire photographic images of the subject; obtaining a topography map representative of a three-dimensional (3D) surface geometry of the subject using a 3D mapping assembly of the imaging system; developing a scanning plan based on the topography map, the scanning plan comprising a traversal path of the image acquisition assembly and a camera triggering order along the traversal path, the traversal path maintaining a nominal working distance between the subject and the camera and further maintaining optical axis orientation near normal to skin FOV during each image capture; executing the scanning plan via an actuator of the imaging system, the location sensor, and the image acquisition assembly to obtain significantly overlapping images along the traversal path; deblurring each captured image utilizing the respective section of the topography map; and combining the overlapping images taken by the image acquisition assembly into a sequence of high-resolution dermatoscopy and skin images.

20. The method of claim 19, wherein at each image capture event, skin and dermatoscopy images of a near similar FOV are captured near-simultaneously on separate images, and wherein the illumination assembly is configured to emit a selected spectrum and intensity for each image type.

Images