Removing background noise from an image

Abstract

A computer-implemented method for removing background noise from a medical image comprising voxels, each voxel having a voxel intensity, the method comprising: generating a mask based on the medical image, the mask including a foreground portion designating foreground voxels, a background portion designating background voxels, and a perimeter separating the foreground portion from the background portion; specifying a perimeter portion of the mask, the perimeter portion surrounding the perimeter, a subset of the foreground portion, and a subset of the background portion; specifying a threshold for the perimeter portion to separate voxels based on voxel intensity; filtering the medical image with the mask and the threshold to obtain a filtered image; and displaying the filtered image on a display.

Description

[Technical Field]

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS This application claims priority to U.S. Provisional Application No. 63 / 411,485, filed September 29, 2022, and U.S. Patent Application No. 18 / 373,735, filed September 27, 2023, the contents of which are incorporated by reference herein in their entireties. [Background technology]

[0002] Tumor treatment fields (TT fields) are low-intensity alternating current electric fields in the mid-frequency range (e.g., 50 kHz to 1 MHz) that can be used to treat tumors, as described in U.S. Patent No. 7,565,205. TT fields are noninvasively induced in target regions by placing transducers directly on the subject's body and applying an alternating current (AC) voltage between the transducers. Conventionally, a first pair of transducers and a second pair of transducers are placed on the subject's body. An AC voltage is applied between the first pair of transducers for a first time interval, generating an electric field with field lines running generally in the anterior-posterior direction. Then, an AC voltage of the same frequency is applied between the second pair of transducers for a second time interval, generating an electric field with field lines running generally in the lateral direction. The system repeats this two-step sequence throughout the treatment.

[0003] TT field treatment planning involves segmenting tissue from background voxels on medical images (e.g., magnetic resonance imaging (MRI) images) to determine where to place transducers on a subject's body and evaluate the distribution of TT fields and quantitative treatment effects. Manual segmentation is time-consuming, and traditional computer-implemented segmentation can lack accuracy. Furthermore, traditional approaches can suffer from large amounts of data and data annotation, which can lead to labeling noise and intra- and inter-observer variability. [Brief explanation of the drawings]

[0004] [Figure 1]1 is a flowchart of an example of removing background from a medical image. [Figure 2] 1 is a flowchart of an example of filtering a medical image using a mask. [Figure 3A] An example of defining the outer periphery is shown below. [Figure 3B] An example of defining the outer periphery is shown below. [Figure 3C] An example of defining the outer periphery is shown below. [Figure 3D] An example of defining the outer periphery is shown below. [Figure 4A] An example of a medical image is shown. [Figure 4B] An example of a medical image is shown. [Figure 4C] An example of a medical image is shown. [Figure 5A] 1 shows an example of a histogram-adjusted medical image showing existing background noise. [Figure 5B] 1 shows an example of a histogram-adjusted medical image showing existing background noise. [Figure 5C] 1 shows an example of a histogram-adjusted medical image showing existing background noise. [Figure 6A] 1 shows an example of a filtered medical image. [Figure 6B] 1 shows an example of a filtered medical image. [Figure 6C] 1 shows an example of a filtered medical image. [Figure 7A] 1 shows an example of a filtered medical image. [Figure 7B] 1 shows an example of a filtered medical image. [Figure 7C] 1 shows an example of a filtered medical image. [Figure 8A] 1 shows an example of a medical image filtered at a high discrete level. [Figure 8B] 1 shows an example of a medical image filtered at a high discrete level. [Figure 8C] 1 shows an example of a medical image filtered at a high discrete level. [Figure 9A] 1 shows an example of a medical image filtered at discrete levels near the midpoint of the discrete levels. [Figure 9B] 1 shows an example of a medical image filtered at discrete levels near the midpoint of the discrete levels. [Figure 9C] 1 shows an example of a medical image filtered at discrete levels near the midpoint of the discrete levels. [Figure 10A] 1 shows an example of a medical image filtered at a low discrete level. [Figure 10B] 1 shows an example of a medical image filtered at a low discrete level. [Figure 10C] 1 shows an example of a medical image filtered at a low discrete level. [Figure 11] 1 shows an example of a filtered image generated at multiple discrete levels. [Figure 12A] 1 shows an example of a visualization of an exemplary mask with different discrete levels. [Figure 12B] 1 shows an example of a visualization of an exemplary mask with different discrete levels. [Figure 12C] 1 shows an example of a visualization of an exemplary mask with different discrete levels. [Figure 13A] An example of a medical image is shown. [Figure 13B] 1 shows an example of a medical image after background removal. [Figure 14] 1 illustrates an example of a computer device for use in embodiments of the present invention.

[0005] Various embodiments will now be described in detail with reference to the accompanying drawings, in which like reference numerals represent like elements, and in which: DETAILED DESCRIPTION OF THE INVENTION

[0006] To provide effective TT field therapy to a subject, precise locations for transducer placement on the subject's body must be generated, and these precise locations are based on, for example, the type of cancer, the size of the cancer, and the location of the cancer in the subject's body. However, determining these precise locations is difficult, and this determination is typically made by computer simulation of numerous possible transducer placement locations.

[0007] Such computer simulations are constructed from images (e.g., magnetic resonance imaging (MRI), computed tomography (CT), etc.) used to model the subject. To run the simulation, the computer must perform image segmentation to identify and remove the target tissue from background noise. One challenge in such image segmentation is how to accurately separate foreground voxels (e.g., target tissue voxels) from background voxels (e.g., noise and artifacts). Another challenge is how to consider user input to improve the accuracy and personalization of the segmentation for a subject. For example, because the size, texture, and shape of the target tissue (e.g., skin) can vary from subject to subject, it can be difficult to identify the perimeter of the target tissue from surrounding background voxels.

[0008] The present inventors have recognized these problems and have discovered a method for generating a mask based on a medical image, and using the mask and a threshold to filter the medical image and remove background from the medical image. By setting a user-adjustable mask perimeter and a user-adjustable threshold, the accuracy and efficiency of background removal can be improved, and the accuracy and efficiency of tumor treatment planning can be improved.

[0009] FIG. 1 is a flowchart illustrating an example of a computer-implemented method 100 for removing background from a medical image. In some embodiments, the image is not limited to a medical image and may be any type of image. Certain steps of method 100 are described as steps performed by a computer. A computer is any device including one or more processors and memory accessible by the one or more processors, where the memory stores instructions that, when executed by the one or more processors, cause the computer to perform the relevant steps of method 100. While FIG. 1 depicts an order of operations for purposes of illustration, the timing and order of such operations may be changed where appropriate without negating the purpose and advantages of the examples detailed throughout this disclosure.

[0010] 1 , in step 102, the method may include acquiring medical images having voxels. The medical images may include, for example, at least one of a magnetic resonance imaging (MRI) image, a computed tomography (CT) image, an X-ray image, an ultrasound image, a nuclear medicine image, a positron emission tomography (PET) image, an arthrography image, a myelography image, or any image of a subject's body that provides an internal view of the subject's body. Each image may include an outline of a portion of the subject's body and a region corresponding to a region of interest (e.g., a tumor) within the subject's body. In one example, the medical images may be three-dimensional (3D) MRI images.

[0011] In step 104, method 100 may include performing one or more preprocessing procedures on the medical image. In some embodiments, the preprocessing procedure may include at least one of Gaussian smoothing or bias correction. In one example, the bias correction is N4 bias correction. In some embodiments, the preprocessing procedure produces a smoother, less biased image. In some embodiments, the preprocessing procedure may include scaling voxel intensities of voxels in the medical image to obtain a scaled medical image. In one example, the following steps 106 through 118 are performed on the scaled medical image.

[0012] In step 106, method 100 may include generating a mask based on the medical image. In some embodiments, the mask includes a foreground portion that designates foreground voxels, a background portion that designates background voxels, and a perimeter that separates the foreground and background portions. In some embodiments, the mask is generated through at least one of a multi-Otsu thresholding process, a k-means clustering process, or a morphological segmentation process.

[0013] In some embodiments, the foreground portion having foreground voxels represents a desired tissue (e.g., a target tissue) in the medical image, and the background portion having background voxels represents one or more regions in the medical image that are devoid of the desired tissue. In some embodiments, the foreground portion represents a region of interest in the medical image, and the background portion represents one or more regions in the medical image that are outside the region of interest. In one example, the desired tissue is at least one of skin, bone, skull, organ, brain, or any tissue of the human body. In a more specific example, the foreground portion corresponds to the subject's head. As another example, the foreground portion corresponds to the subject's torso.

[0014] In step 108, method 100 may include specifying a perimeter portion of the mask, a perimeter portion surrounding the perimeter, a subset of the foreground portion, and a subset of the background portion. In some embodiments, the perimeter portion includes a foreground perimeter separating the perimeter portion from the remainder of the foreground portion and a remainder of the foreground portion that does not include the subset of the foreground portion. In some embodiments, the perimeter portion further includes a background perimeter separating the perimeter portion from the remainder of the background portion and a remainder of the background portion that does not include the subset of the background portion. In some embodiments, the perimeter is approximately equidistant from the foreground perimeter and the background perimeter. In some embodiments, the width of the perimeter portion between the foreground perimeter and the background perimeter is approximately 5 mm, 10 mm, 15 mm, 20 mm, or 25 mm. The width of the perimeter portion between the foreground perimeter and the background perimeter may vary depending on the subject and imaging parameters. Examples of specifying a perimeter portion of the mask are shown in FIGS. 4A-4C.

[0015] In step 110, method 100 may include specifying a perimeter threshold to separate voxels based on voxel intensity. In some embodiments, the threshold is a weighted threshold. In some embodiments, the threshold is user-defined. As an example, the threshold may be a user-defined weighted threshold.

[0016] In step 112, method 100 may include filtering the medical image using the mask to obtain a filtered image. In some embodiments, filtering the medical image is based on voxel intensity. For example, for voxels in the medical image that are located in the remaining foreground portion of the mask, their voxel intensity is maintained in the filtered image. Voxels in the medical image that are located in the remaining background portion of the mask are assigned a background voxel intensity. In some embodiments, for voxels in the medical image that are located in the periphery of the mask, a voxel intensity is assigned in the filtered image based on a threshold value and the voxel's location in the periphery. For example, if the voxel intensity of a particular voxel is above the threshold specified in step 110, the voxel intensity of the particular voxel is maintained in the filtered image; if the voxel intensity of the particular voxel is below the threshold, the voxel intensity of the particular voxel is assigned a background voxel intensity. An example of filtering a medical image using a mask is shown in FIG. 2, which is further described below.

[0017] In some embodiments, for voxels located in the perimeter portion within the medical image, the medical image may be filtered based on the voxel intensity and the distance of the voxel from the perimeter portion. In one example, filtering the medical image may include averaging the voxel intensity of the voxel and an intensity assigned to the voxel based on the distance of the voxel from the perimeter portion. In one example, filtering the medical image may include assigning an intensity to the voxel based on its distance relative to a location in the foreground portion. In other words, the voxel intensity of a corresponding location within the perimeter portion may be based on the distance of the location from the foreground perimeter of the perimeter portion or the background perimeter of the perimeter portion. In some embodiments, the intensity assigned to the voxel is related to the distance between the voxel and a location in the foreground portion. In some embodiments, the intensity assigned to the voxel is proportional to the distance between the voxel and a location in the foreground portion. In some embodiments, the intensity assigned to the voxel is proportional to the distance between the voxel and the perimeter of the perimeter portion. In some embodiments, the intensity assigned to the voxel is proportional to the distance between the voxel and a location on the foreground perimeter of the perimeter portion. As an example, the smaller the distance to the location of the foreground portion, the greater the intensity assigned to the voxel. As an example, the intensity assigned to the voxel may be the average of the voxel intensity and an intensity assigned to the voxel based on the distance of the voxel to the location of the foreground portion, and this assigned intensity may be represented by the following formula:

[0018] p レベル =(m1+m2) / 2 Equation (1) where p レベル is the voxel intensity assigned to the voxel, m1 is the voxel intensity of the voxel in the medical image, and m2 is the intensity assigned to the voxel in proportion to the distance between the voxel and the location of the foreground part.

[0019] In some embodiments, for voxels located in the periphery of the medical image, filtering the medical image comprises a weighted averaging of the voxel's voxel intensity and an intensity assigned to the voxel based on the voxel's distance to the periphery. In other words, voxels in the medical image located in the periphery of the mask are assigned a voxel intensity in the filtered image based on a weighted combination of the voxel's voxel intensity in the medical image and another voxel intensity based on the voxel's location in the periphery. In some embodiments, for voxels located in the periphery of the medical image, filtering the medical image comprises a voxel-by-voxel weighted sum of the voxel's normalized voxel intensity in the medical image and the voxel intensity of a corresponding location in the periphery of the mask. As an example, the filtered image is generated based on the following formula:

[0020] p レベル =w1*m1+w2*m2,w2=1-w1 formula (2) where m1 is the voxel intensity of a voxel in the medical image, m2 is the voxel intensity of the corresponding location in the perimeter of the mask, and w1 and w2 are weighting parameters.

[0021] In some embodiments, the threshold value specified in step 110 is a weighted threshold value. As an example, if voxel intensities located in the peripheral portion within the medical image exceed the weighted threshold value, filtering the medical image includes assigning intensities to voxels based on voxel intensities in the medical image or the preprocessed medical image. As an example, if voxel intensities located in the peripheral portion within the medical image exceed the weighted threshold value, filtering the medical image includes assigning intensities to voxels based on a weighted average of the intensities. In some embodiments, the threshold value is a user-defined weighted threshold value. In one example, method 100 may include filtering the medical image using the mask and the user-defined weighted threshold value to obtain a filtered image.

[0022] In step 114, method 100 may include performing one or more post-processing procedures on the filtered image. In some embodiments, the post-processing procedures include at least one of a morphological closing operation, a three-dimensional (3D) hole filling operation, or a smoothing operation. Step 114 may be performed optionally.

[0023] In step 116, method 100 may include displaying the filtered image. As an example, the filtered image generated in step 112 is displayed. As an example, the filtered image generated in step 114 is displayed. In some embodiments, the method further includes displaying voxel intensities of the medical image and / or displaying a weighted average voxel intensity of voxels based on their distance from a perimeter.

[0024] In step 118, method 100 may include adjusting the size and / or threshold of the perimeter. In some embodiments, the method includes receiving user input for adjusting filtering parameters to obtain adjusted filtering parameters, the adjusted filtering parameters including at least one of an adjusted size of the perimeter or an adjusted threshold for the perimeter. As an example, the method includes adjusting the size of the perimeter to obtain a modified mask. As an example, the method includes adjusting the threshold to obtain a modified threshold. As an example, the threshold is a weighted threshold. As described in step 112 above, for voxels in the medical image located on the perimeter of the mask, a voxel intensity is assigned in the filtered image based on the threshold and the voxel's position in the perimeter. For example, if the voxel intensity is above the threshold, the voxel intensity is maintained in the filtered image; if the voxel intensity is below the threshold, the assigned voxel intensity becomes the background voxel intensity. Thus, a user can adjust the threshold to adjust the perimeter (e.g., the perimeter of the skin) of the filtered image. As an example, a user may be able to adjust the threshold and see the modified filtered image in real time. As an example, the method may include adjusting the size of the perimeter to obtain a modified mask and adjusting the threshold to obtain a modified threshold.

[0025] In some embodiments, one or more users (e.g., physicians, nurses, assistants, staff, physicists, dosimetrists, etc.) may use a user interface to adjust the perimeter portion and / or threshold of the mask using user-adjustable levels. In some embodiments, the user-adjustable levels are interactive sliders in the user interface, which define an ablation region proximate the outer surface (e.g., perimeter) of the tissue in the medical image. As an example, the interactive slider has discrete levels that are user-adjustable.

[0026] In some embodiments, an interactive slider (or other user interface) may be user adjustable to adjust the discrete levels or select the level of the perimeter within a predetermined range of discrete levels. In some embodiments, the user-adjustable discrete levels have a default setting. As an example, the user-adjustable discrete levels may be user-adjustable from the default setting to increase the size of the perimeter, or may be user-adjustable from the default setting to decrease the size of the perimeter. For example, the default setting is the perimeter of the mask that separates foreground and background voxels. As an example, the user-adjustable discrete levels may have from about 8 to about 512 discrete levels, with the default user-adjustable levels being approximately the middle of the discrete levels (e.g., the perimeter of the mask). As an example, the user-adjustable discrete levels may have 255 discrete levels, with the default user-adjustable level being 126. In this example, the value 126 defines the perimeter of the mask that separates foreground and background voxels. Examples of user-adjustable discrete levels are shown in FIG. 3D, which is described in more detail below. Examples of these embodiments of user-defined perimeters are shown in FIG. 12, which is described in more detail below.

[0027] After the filtering parameters are adjusted, the flow of method 100 proceeds to step 112 and the loop is repeated. In some embodiments, the method further includes filtering the medical image using the modified mask and / or the modified threshold to obtain a modified filtered image and displaying the modified filtered image on a display.

[0028] In some embodiments, method 100 further includes generating and outputting, based on the filtered image, one or more recommendations of locations on the subject's body for placing one or more transducers for applying tumor treating fields to the subject's body. In some embodiments, the selection of locations for placing the one or more transducers may be further based on a region of interest on the subject's body that corresponds to, for example, a tumor.

[0029] Figure 2 is a flowchart of an example of filtering a medical image using a mask. Certain steps of method 200 are described as computer-implemented steps. A computer is any device that includes one or more processors and memory accessible by the one or more processors, where the memory stores instructions that, when executed by the one or more processors, cause the computer to perform the relevant steps of method 200. While an order of operations is shown in Figure 2 for purposes of explanation, the timing and order of such operations may be changed where appropriate without negating the purpose and advantages of the examples detailed throughout this disclosure.

[0030] In step 202, for the remaining voxels in the foreground portion, filtering the medical image includes preserving the voxel intensity of the voxels. In step 204, for the remaining voxels in the background portion, filtering the medical image includes assigning the voxels a background voxel intensity.

[0031] For voxels in the outer perimeter, if the voxel intensity is above the threshold, the voxel intensity of the voxel is maintained in step 206, and if the voxel intensity is below the threshold, the voxel is assigned the background voxel intensity in step 208. In one example, the threshold is a weighted threshold.

[0032] In some embodiments, if the voxel intensity is above a threshold as in step 206, the voxel intensity is set to the same as the voxel intensity in the medical image. In other words, if the voxel intensity is above a threshold, the voxel intensity in the medical image is maintained. In some embodiments, if the voxel intensity is above a threshold, the voxel intensity is set to the voxel intensity in the image after the post-processing procedure. In some embodiments, if the voxel intensity is above a threshold, the voxel intensity is set to be the average of the voxel intensity in the medical image and an intensity based on the distance from the voxel's perimeter as calculated by equation (1) above. In some embodiments, if the voxel intensity is above a threshold, the voxel intensity is set to be a weighted average of the voxel intensity in the medical image and an intensity based on the distance from the voxel's perimeter as calculated by equation (2) above.

[0033] 3A-3D show an example of defining a perimeter. FIG. 3A shows a mask image generated based on a medical image or a preprocessed medical image. In this example, the mask is composed of a foreground portion 302 that specifies foreground voxels, a background portion 304 that specifies background voxels, and a perimeter 306 that separates the foreground and background portions. Foreground portion 302 with foreground voxels represents a desired tissue in the medical image, and background portion 304 with background voxels represents one or more regions in the medical image that are devoid of the desired tissue. In the example shown in FIG. 3A, foreground portion 302 represents the subject's head.

[0034] 3B and 3C illustrate an example of specifying a perimeter portion 312 of a mask. The perimeter portion 312 surrounds the perimeter 306, a subset of the foreground portion (e.g., the portion between the foreground perimeter 308 and the perimeter 306), and a subset of the background portion (e.g., the portion between the background perimeter 310 and the perimeter 306). The perimeter portion 312 includes the foreground perimeter 308, which separates the perimeter portion 312 from the remainder of the foreground portion, but the remainder of the foreground portion does not include the subset of the foreground portion. Furthermore, the perimeter portion 312 further includes the background perimeter 310, which separates the perimeter portion 312 from the remainder of the background portion, but the remainder of the background portion does not include the subset of the background portion. In one example, the perimeter 306 is approximately equidistant from the foreground perimeter 308 and the background perimeter 310. As an example, the width of the perimeter 312 between the foreground perimeter 308 and the background perimeter 310 is approximately 5 mm, 10 mm, 15 mm, 20 mm, or 25 mm. In some embodiments, the perimeter of the mask is defined by the user. As an example, a user interface with an interactive slider for specifying the perimeter may be provided.

[0035] 3D shows an example of adjusting the perimeter portion using a user interface 318 to adjust the discrete levels of the perimeter portion. In this example, the user interface 318 includes several discrete levels 320 that a user can select from and a slider to assist the user in selecting the discrete level. Thus, a user may use the slider to increase or decrease the size of the perimeter portion. As an example, the lowest setting of discrete levels may correspond to the background perimeter 310, and the highest setting of discrete levels may correspond to the foreground perimeter 308.

[0036] As an example, adjusting slider 318 moves both the foreground perimeter 308 and the background perimeter 310. In this example, slider 318 has two slides: a first slide 322 moves the background perimeter 310, and a second slide 324 moves the foreground perimeter 308. In the position of the first slide 322 shown in Figure 3D, the background perimeter 310 is moved to position 316, and in the position of the second slide 324 shown in Figure 3D, the foreground perimeter 308 is moved to position 314. Thus, the perimeter portions are adjusted between positions 316 and 314.

[0037] As an example, adjusting slider 318 moves background perimeter 310 while keeping foreground perimeter 308 constant. In this example, slider 318 may include only one slide, first slide 322, for moving background perimeter 310. In the position of first slide 322 shown in FIG. 3D , background perimeter 310 is moved to position 316 while foreground perimeter 308 is not moved. Thus, the perimeter portion is adjusted between positions 316 and 308.

[0038] As an example, adjusting slider 318 moves foreground perimeter 308 while background perimeter 310 remains constant. In this example, slider 318 has only one slide: second slide 324 for moving foreground perimeter 308. In the position of second slide 324 shown in Figure 3D, foreground perimeter 308 is moved to position 314 while background perimeter 310 is not moved. In this manner, the perimeter portion is adjusted between positions 314 and 310.

[0039] 4A to 4C show examples of medical images. In these examples, the medical images are MRI images of a subject's head. FIG. 4A shows an MRI image of the subject's head at an axial angle, FIG. 4B shows an MRI image of the subject's head at a sagittal angle, and FIG. 4C shows an MRI image of the subject's head at a coronal angle.

[0040] 5A-5C show the example medical images shown in FIGS. 4A-4C with histogram adjustment to show the existing background noise.

[0041] 6A-6C illustrate examples of the medical images shown in FIGS. 4A-4C that have been filtered in accordance with an exemplary embodiment of the present invention. In the examples shown in FIGS. 6A-6C, the medical images were filtered using a mask that includes a perimeter that separates voxels considered to be foreground voxels from voxels considered to be background voxels. When filtered with the mask, voxels considered to be foreground voxels retain their values, while voxels considered to be background voxels were assigned a constant background value (here, "0," which corresponds to black). Thus, in these filtered medical images, foreground voxels are depicted in a retained gray color, and background voxels are depicted in black. The mask was generated in accordance with step 106 of FIG. 1.

[0042] 7A-7C show another example of the medical image shown in FIGS. 4A-4C filtered in accordance with an exemplary embodiment of the present invention. Similar to the example shown in FIGS. 6A-6C, the medical image was filtered using a mask including a perimeter separating voxels considered to be foreground voxels from voxels considered to be background voxels. When filtered with the mask, voxels considered to be foreground voxels maintained their values, while voxels within the perimeter that had a background color (e.g., "0" for black) were assigned a value not associated with a background color (e.g., a non-zero value that is not black). Voxels considered to be background voxels were assigned a constant background value (here, "0," corresponding to black). Thus, in these filtered medical images, foreground voxels are rendered with a maintained gray color or adjusted to avoid becoming background, and background voxels are rendered with a black color. The mask is generated in accordance with step 106 of FIG. 1.

[0043] 8A-10C show examples of filtered medical images with various user-adjustable discrete levels used to adjust the size of the mask's outer perimeter. In these examples, foreground voxels of a subject's head at various angles are shown in gray, while background voxels are filtered and shown in black. In the example shown in FIGS. 8A-10C, there are 255 user-adjustable discrete levels. The user-adjustable levels adjust the size of the mask's outer perimeter. Decreasing the discrete level increases the size of the outer perimeter, while increasing the discrete level decreases the size of the outer perimeter. In this example, the background perimeter of the outer perimeter (similar to background perimeter 308 in FIG. 3D ) is kept constant, while the foreground perimeter (similar to foreground perimeter 310 in FIG. 3D ) is user-adjustable across 255 discrete levels, with level 255 being closest to the background perimeter and level 1 being farthest from the background perimeter.

[0044] 8A-8C show an example of a medical image filtered at a high discrete level. In this example, the discrete level of the generated filtered image is 223. As shown in FIGS. 8A-8C, the foreground voxels (gray) have been reduced in size, and the background voxels (black) have encroached on the periphery of the foreground voxels, as shown in portion 801.

[0045] 9A-9C show an example of a medical image filtered at a discrete level near the midpoint of the discrete levels. In this example, the discrete level of the generated filtered image is 172. As shown in FIGS. 9A-9C, the foreground voxels (gray) have clear outlines, and the background voxels (block color) have been removed with little error.

[0046] 10A-10C show an example of a medical image filtered at a low discrete level. In this example, the discrete level of the generated filtered image is 18. As shown in FIGS. 10A-10C, the size of the foreground voxels (gray) is increased so that the outer periphery of the foreground voxels includes unwanted background voxels, such as portion 1001.

[0047] FIG. 11 shows an example of a generated filtered image having multiple discrete levels. In the example shown in FIG. 11, the foreground voxels include the subject's head and voxels on the periphery of the subject's head. The periphery is the region of the subject's head from the lowest discrete level to the highest discrete level. The periphery closer to the inside of the subject's head has higher discrete levels, while the periphery closer to the background voxels has lower discrete levels. The discrete levels 1101 within the periphery are shown in different shades of gray.

[0048] 12A-12C show an example of visualization of an exemplary mask with different discrete levels. In the example shown in FIGS. 12A-12C, the masked portions of voxels considered to be foreground voxels are shown in the same gray color, and the masked portions of voxels considered to be background voxels are shown in the background color (here, black). As shown in FIGS. 12A-12C, the discrete levels of the peripheral portions of the mask capture different features of the subject, as indicated by the varying gray color values.

[0049] 13A shows an example medical image, and FIG. 13B shows an example medical image after background removal with a mask in accordance with an exemplary embodiment. In this example, the mask includes 126 discrete levels out of 255 discrete levels. As noted above, the resulting filtered medical image clearly shows the circumference of the subject's head.

[0050] Figure 14 illustrates an example of a computing device for use in embodiments of the present invention. As an example, device 1400 may be a computer that implements certain inventive techniques disclosed herein, such as removing background noise from medical images. For example, the methods of Figures 1 and 2 may be performed by a computer such as device 1400. Device 1400 may include one or more processors 1402, memory 1403, one or more input devices, and one or more output devices 1405.

[0051] In one example, based on input 1401, one or more processors remove background from an image according to embodiments herein. In one example, input 1401 is a user input. In another example, input 1401 may be from another computer in communication with apparatus 1400. Input 1401 may be received in combination with one or more input devices (not shown) of apparatus 1400.

[0052] The memory 1403 is accessible by one or more processors 1402 (e.g., via link 1404), and the one or more processors 1402 may read information from and write information to the memory 1403. The memory 1403 may store instructions that, when executed by the one or more processors 1402, implement one or more embodiments described herein. The memory 1403 may be a non-transitory computer-readable medium (or non-transitory processor-readable medium) that includes a set of instructions for removing background noise from medical images, which, when executed by a processor (such as one or more processors 1402), cause the processor to perform one or more methods disclosed herein.

[0053] The one or more output devices 1405 may provide a status of the computer-implemented techniques herein. The one or more output devices 1405 may provide visualization data such as medical images, masks, filtered images, and / or voxel intensities of medical images according to certain embodiments of the present invention. The one or more output devices 1405 may display user-adjustable levels, which may be controlled using the input 1401.

[0054] Device 1400 may be an apparatus for removing background noise from medical images and includes one or more processors (e.g., one or more processors 1402) and memory accessible by the one or more processors (e.g., memory 1403), which stores instructions that, when executed by the one or more processors, cause the apparatus to perform one or more methods disclosed herein. Illustrative Embodiments

[0055] The present invention also includes other exemplary embodiments such as the following.

[0056] Exemplary Embodiment 1. A computer-implemented method for removing background noise from a medical image comprising voxels, each voxel having a voxel intensity, the method comprising: generating a mask based on the medical image, the mask including a foreground portion designating foreground voxels, a background portion designating background voxels, and a perimeter separating the foreground portion from the background portion; specifying a perimeter portion of the mask, the perimeter portion surrounding the perimeter, a subset of the foreground portion, and a subset of the background portion; specifying a threshold value for the perimeter portion to separate voxels based on voxel intensity; filtering the medical image using the mask and the threshold value to obtain a filtered image; and displaying the filtered image on a display.

[0057] Exemplary Embodiment 2. The method of embodiment 1, wherein after the step of filtering the medical image using the mask, voxels in the medical image that are located in the remaining foreground portion of the mask maintain their voxel intensity in the filtered image, voxels in the medical image that are located in the remaining background portion of the mask are assigned a background voxel intensity, voxels in the medical image that are located in the outer portion of the mask maintain their voxel intensity in the filtered image if their voxel intensity is above a threshold, and voxels in the medical image that are located in the outer portion of the mask are assigned a background voxel intensity in the filtered image if their voxel intensity is below the threshold.

[0058] Exemplary Embodiment 3. The method of exemplary embodiment 1, wherein voxels in the medical image that lie within the perimeter of the mask are assigned a voxel intensity in the filtered image based on a threshold value and the location of the voxel within the perimeter.

[0059] Exemplary Embodiment 4. The method of exemplary embodiment 1, wherein for voxels located in the outer periphery within the medical image, the step of filtering the medical image includes averaging the voxel intensity of the voxel and an intensity assigned to the voxel based on the voxel's distance from the outer periphery.

[0060] Exemplary Embodiment 5. The method of exemplary embodiment 1, wherein for voxels located in the outer periphery within the medical image, the step of filtering the medical image includes assigning an intensity to the voxel based on its distance relative to a position within the foreground portion.

[0061] Exemplary Embodiment 6 The method of exemplary embodiment 5, wherein the smaller the distance to the location of the foreground portion, the greater the intensity assigned to the voxel.

[0062] Exemplary Embodiment 7. The method of exemplary embodiment 1, wherein for voxels located in the outer periphery within the medical image, the step of filtering the medical image includes weighted averaging the voxel intensity of the voxel and an intensity assigned to the voxel based on the voxel's distance from the outer periphery.

[0063] Exemplary Embodiment 8. The method of exemplary embodiment 1, wherein voxels in the medical image located in the peripheral portion of the mask are assigned a voxel intensity in the filtered image based on a weighted combination of the voxel intensity of the voxel in the medical image and another voxel intensity based on the voxel's position in the peripheral portion.

[0064] Exemplary Embodiment 9. The method of exemplary embodiment 1, wherein for voxels located in a peripheral portion within the medical image, the step of filtering the medical image includes a voxel-by-voxel weighted sum of the voxel's normalized voxel intensity and the voxel intensity of a corresponding location within the peripheral portion.

[0065] Exemplary Embodiment 10. The method of embodiment 8, wherein the perimeter portion includes a foreground perimeter separating the perimeter portion from the remainder of the foreground portion, and the perimeter portion includes a background perimeter separating the perimeter portion from the remainder of the background portion, and wherein voxel intensities at corresponding locations within the perimeter portion are based on the distance of the location from the foreground perimeter of the perimeter portion or the background perimeter of the perimeter portion.

[0066] Exemplary Embodiment 11. The method of embodiment 1, comprising: scaling voxel intensities of voxels in the medical image to obtain a scaled medical image; and assigning scaled voxel intensities to voxels in a perimeter portion of the mask based on a distance of the voxel in the mask from a foreground perimeter separating the perimeter portion from the remainder of the foreground portion or a background perimeter separating the perimeter portion from the remainder of the background portion; wherein filtering the medical image using the mask comprises filtering the scaled medical image using the mask and a threshold; and for voxels in the scaled medical image located in the remainder of the foreground portion of the mask, filtering maintaining voxel intensities of voxels in the filtered image; and for voxels in the scaled medical image that lie in the remainder of the background portion of the mask, assigning background voxel intensities to the voxels in the filtered image; and for voxels in the scaled medical image that lie in the outer portion of the mask, assigning voxel intensities to the voxels in the filtered image based on a weighted combination of the scaled voxel intensities of the scaled medical image and the scaled voxel intensities of the voxels in the outer portion of the mask if the voxel intensity is above a threshold, and assigning background voxel intensities to the voxels in the filtered image if the voxel intensity is below the threshold.

[0067] Exemplary Embodiment 12. The method of exemplary embodiment 1, wherein the threshold is a user-defined weighted threshold, and wherein filtering the medical image includes filtering the medical image using the mask and the user-defined weighted threshold to obtain a filtered image.

[0068] Exemplary Embodiment 13. The method of exemplary embodiment 1, wherein the threshold is a weighted threshold, and for voxels in the medical image located in the outer periphery having an intensity above the weighted threshold, the step of filtering the medical image includes assigning an intensity to the voxel based on the voxel intensity in the medical image.

[0069] Exemplary Embodiment 14. The method of exemplary embodiment 1, further comprising preprocessing the medical image prior to generating the mask to obtain a preprocessed medical image, wherein the threshold is a weighted threshold, and for voxels in the medical image located in the outer periphery having intensities above the weighted threshold, filtering the medical image comprises assigning intensities to the voxels based on the voxel intensities in the preprocessed medical image.

[0070] Exemplary Embodiment 15. The method of exemplary embodiment 1, wherein the threshold is a weighted threshold, and the step of filtering the medical image for voxels in the medical image located in the outer periphery having an intensity above the weighted threshold includes assigning an intensity to the voxel based on a weighted average of the intensities.

[0071] Exemplary Embodiment 16. The method of embodiment 1, wherein the perimeter portion includes a foreground perimeter separating the perimeter portion from the remainder of the foreground portion, where the remainder of the foreground portion does not include the subset of the foreground portion, and a background perimeter separating the perimeter portion from the remainder of the background portion, where the remainder of the background portion does not include the subset of the background portion.

[0072] Exemplary Embodiment 17. The method of exemplary embodiment 16, wherein the perimeter is approximately equidistant from the foreground perimeter and the background perimeter.

[0073] Exemplary Embodiment 18 The method of embodiment 16, wherein the width of the perimeter portion between the foreground perimeter and the background perimeter is about 10 mm.

[0074] Exemplary Embodiment 19. The method of exemplary embodiment 1, wherein the perimeter of the mask is user-defined and the perimeter threshold is user-defined.

[0075] Exemplary Embodiment 20. The method of embodiment 1, further comprising providing a user interface with an interactive slider for specifying the perimeter portion.

[0076] Exemplary Embodiment 21. The method of exemplary embodiment 1, further comprising the step of displaying voxel intensities of the medical image and / or displaying a weighted average voxel intensity of voxels based on their distance from a perimeter of the voxels.

[0077] Exemplary Embodiment 22. The method of exemplary embodiment 1, wherein after the step of displaying the image, the method further includes the steps of receiving user input for adjusting filtering parameters to obtain adjusted filtering parameters, wherein the adjusted filtering parameters include at least one of an adjusted size of the peripheral portion or an adjusted threshold value of the peripheral portion; filtering the medical image with the adjusted filtering parameters to obtain a modified filtered image; and displaying the modified filtered image on a display.

[0078] Exemplary Embodiment 23. The method of exemplary embodiment 1, wherein after the step of displaying the image, the method further includes the steps of adjusting the size of the perimeter portion to obtain a modified mask, filtering the medical image using the modified mask and a threshold value to obtain a modified filtered image, and displaying the modified filtered image on a display.

[0079] Exemplary Embodiment 24. The method of exemplary embodiment 1, further comprising, after the step of displaying the image, adjusting the threshold to obtain a modified threshold, filtering the medical image using the mask and the modified threshold to obtain a modified filtered image, and displaying the modified filtered image on a display.

[0080] Exemplary Embodiment 25. The method of exemplary embodiment 1, wherein after the step of displaying the image, the method further includes the steps of adjusting a size of the perimeter portion to obtain a modified mask, adjusting a threshold to obtain a modified threshold, filtering the medical image using the modified mask and the modified threshold to obtain a modified filtered image, and displaying the modified filtered image on a display.

[0081] Exemplary Embodiment 26. The method of embodiment 1, wherein the foreground portion represents a desired tissue within the medical image and the background portion represents one or more regions within the medical image that are devoid of the desired tissue.

[0082] Exemplary Embodiment 27. The method of embodiment 1, wherein the foreground portion represents a region of interest in the medical image and the background portion represents one or more regions in the medical image that are not included in the region of interest.

[0083] Exemplary Embodiment 28 The method of exemplary embodiment 1, wherein the mask is generated by at least one of a multi-Otsu thresholding process, a k-means clustering process, or a morphological segmentation process.

[0084] Exemplary Embodiment 29. The method of exemplary embodiment 1, further comprising performing a post-processing procedure on the filtered image, the post-processing procedure comprising at least one of a morphological closing operation, a three-dimensional (3D) hole-filling operation, or a smoothing operation.

[0085] Exemplary Embodiment 30. The method of exemplary embodiment 1, further comprising performing a pre-processing procedure on the medical image before generating the mask, the pre-processing procedure comprising at least one of Gaussian smoothing or bias correction.

[0086] Exemplary Embodiment 31. The method of exemplary embodiment 1, wherein the medical image includes at least one of a magnetic resonance image, an ultrasound image, a computed tomography image, or an X-ray image.

[0087] Exemplary Embodiment 32. The method of exemplary embodiment 1 further includes generating and outputting, based on the filtered image, one or more recommendations of locations on the subject's body for placing one or more transducers for applying a tumor treatment field to the subject's body.

[0088] Exemplary Embodiment 33. A computer-implemented method for processing medical images, the computer including one or more processors and a memory accessible by the one or more processors, the memory storing instructions that, when executed by the one or more processors, cause the computer to perform the method, the method including: generating a mask based on the medical image, the mask including a foreground portion designating foreground voxels, a background portion designating background voxels, and a perimeter separating the foreground portion from the background portion; specifying a perimeter portion of the mask, the perimeter portion surrounding the perimeter, a subset of the foreground portion, and a subset of the background portion; specifying a threshold for the perimeter portion to separate voxels based on voxel intensity; filtering the medical image using the mask and the threshold to obtain a filtered image; receiving user input to adjust filtering parameters to obtain adjusted filtering parameters, the adjusted filtering parameters including at least one of an adjusted size of the perimeter portion or an adjusted threshold for the perimeter portion; filtering the medical image using the adjusted filtering parameters to obtain a modified filtered image; and displaying the modified filtered image on a display.

[0089] Exemplary embodiment 34. An apparatus for removing background noise from a medical image, comprising: one or more processors; and a memory, wherein the memory has stored therein processor-executable instructions that, when executed by the one or more processors, cause the apparatus to perform the following steps: generating a mask based on the medical image, the mask including a foreground portion that designates foreground voxels, a background portion that designates background voxels, and a perimeter that separates the foreground portion from the background portion; specifying a perimeter portion of the mask, the perimeter portion enclosing the perimeter, a subset of the foreground portion, and a subset of the background portion, the perimeter portion of the mask being user-defined; specifying a threshold for the perimeter portion to separate voxels based on voxel intensity, the threshold for the perimeter portion being user-defined; filtering the medical image using the mask and the threshold to obtain a filtered image; and displaying the filtered image on a display.

[0090] Exemplary Embodiment 35. In the device of exemplary embodiment 34, the device includes a user interface with an interactive slider that defines the perimeter portion.

[0091] Embodiments described under any heading or in any portion of this disclosure may be combined with embodiments described under the same heading or in another portion of this disclosure, unless otherwise stated herein or clearly contradicted by context.

[0092] Numerous modifications, variations, and variations can be made to the described embodiments without departing from the scope of the invention as defined in the claims. The present invention is not intended to be limited to the described embodiments, but rather to have the full scope defined by the language of the following claims and equivalents thereof.

Claims

1. 1. A computer-implemented method for removing background noise from a medical image comprising voxels, each voxel having a voxel intensity, the method comprising: generating a mask based on the medical image, the mask including a foreground portion designating foreground voxels, a background portion designating background voxels, and a perimeter separating the foreground portion and the background portion; specifying a perimeter portion of the mask, a perimeter portion surrounding the perimeter, a subset of the foreground portion, and a subset of the background portion; specifying a perimeter threshold to separate the voxels based on the voxel intensity; filtering the medical image using the mask and the threshold to obtain a filtered image; and displaying the filtered image on a display.

2. After filtering the medical image using the mask, the voxels in the medical image that are located in the remainder of the foreground portion of the mask maintain their voxel intensity in the filtered image; the voxels in the medical image that lie in the remainder of the background portion of the mask are assigned a background voxel intensity; voxels in the medical image located at the outer periphery of the mask maintain their voxel intensity in the filtered image if the voxel intensity is above a threshold; The method of claim 1 , wherein the voxels in the medical image that lie on the periphery of the mask are assigned the background voxel intensity in the filtered image if the voxel intensity is below a threshold.

3. 2. The method of claim 1, wherein for voxels located in a perimeter portion within the medical image, filtering the medical image comprises averaging the voxel intensity of the voxel and an intensity assigned to the voxel based on a distance from the voxel to the perimeter portion.

4. The method of claim 1 , wherein for voxels located in a perimeter portion within the medical image, filtering the medical image includes assigning an intensity to the voxel based on its distance relative to a location within the foreground portion.

5. 2. The method of claim 1, wherein voxels in the medical image located in a perimeter portion of the mask are assigned the voxel intensity in the filtered image based on a weighted combination of the voxel intensity of the voxel in the medical image and another voxel intensity based on the position of the voxel in the perimeter portion.

6. the perimeter portion includes a foreground perimeter separating the perimeter portion from the remainder of the foreground portion; the perimeter portion includes a background perimeter separating the perimeter portion from the remainder of the background portion; The method of claim 5 , wherein voxel intensities at corresponding locations within the perimeter portion are based on the distance of the location from a foreground perimeter of the perimeter portion or a background perimeter of the perimeter portion.

7. scaling the voxel intensities of the voxels in the medical image to obtain a scaled medical image; assigning scaled voxel intensities to voxels in the perimeter portion of the mask based on the distance from the voxels in the mask and from the foreground perimeter separating the perimeter portion from the remainder of the foreground portion or the background perimeter separating the perimeter portion from the remainder of the background portion; filtering the medical image using the mask, filtering the scaled medical image using the mask and the threshold; For voxels in the scaled medical image that are located in the remainder of the foreground portion of the mask, maintaining the voxel intensities of the voxels in the filtered image; For voxels in the scaled medical image that lie in the remaining background portion of the mask, assigning background voxel intensities to the voxels in the filtered image; For voxels of the scaled medical image that are in the outer periphery of the mask, assigning a voxel intensity to a voxel in the filtered image based on a weighted combination of the scaled voxel intensities of the scaled medical image and the scaled voxel intensities of voxels in the outer perimeter of the mask if the voxel intensity is above a threshold; and assigning the background voxel intensity to a voxel in the filtered image if the voxel intensity is below a threshold.

8. the threshold is a user-defined weighted threshold; The method of claim 1 , wherein filtering the medical image comprises filtering the medical image using the mask and the user-defined weighted threshold to obtain a filtered image.

9. the threshold is a weighted threshold; 2. The method of claim 1, wherein if the voxel intensities located in the outer periphery within the medical image are above a weighted threshold, filtering the medical image comprises assigning intensities to the voxels based on the voxel intensities within the medical image.

10. The outer circumferential portion is a foreground perimeter separating the perimeter portion from the remainder of the foreground portion, the remainder of the foreground portion not including the subset of the foreground portion; The method of claim 1 , wherein the perimeter portion comprises a background perimeter separating the perimeter portion from the remainder of the background portion, the remainder of the background portion not including the subset of the background portion.

11. After the step of displaying the image, receiving a user input for adjusting filtering parameters to obtain adjusted filtering parameters, the adjusted filtering parameters including at least one of an adjusted size of the perimeter portion or an adjusted threshold value of the perimeter portion; filtering the medical image using the adjusted filtering parameters to obtain a modified filtered image; The method of claim 1 , further comprising the step of: displaying the modified filtered image on a display.

12. 2. The method of claim 1, comprising: performing a post-processing procedure on the filtered image, the post-processing procedure further comprising at least one of a morphological closing operation, a three-dimensional (3D) hole-filling operation, or a smoothing operation.

13. 10. The method of claim 1, further comprising generating and outputting, based on the filtered image, one or more recommendations of locations on the subject's body for placing one or more transducers for applying a tumor treatment field to the subject's body.

14. 1. A computer-implemented method for processing medical images, the computer including one or more processors and memory accessible by the one or more processors, the memory storing instructions that, when executed by the one or more processors, cause the computer to perform a method, the method comprising: generating a mask based on the medical image, the mask including a foreground portion designating foreground voxels, a background portion designating background voxels, and a perimeter separating the foreground and background portions; specifying a perimeter portion of the mask, a perimeter portion surrounding the perimeter, a subset of foreground portions, and a subset of background portions; thresholding the perimeter to separate the voxels based on voxel intensity; filtering the medical image using the mask and threshold to obtain a filtered image; receiving a user input for adjusting filtering parameters to obtain adjusted filtering parameters, the adjusted filtering parameters including at least one of an adjusted size of the perimeter portion or an adjusted threshold value of the perimeter portion; filtering the medical image using the adjusted filtering parameters to obtain a modified filtered image; and displaying the modified filtered image on a display.

15. 1. An apparatus for removing background noise from a medical image, the apparatus comprising: one or more processors; and a memory, the memory having processor-executable instructions stored therein, the processor-executable instructions, when executed by the one or more processors, performing: generating a mask based on the medical image, the mask including a foreground portion designating foreground voxels, a background portion designating background voxels, and a perimeter separating the foreground and background portions; specifying a perimeter portion of the mask, the perimeter portion surrounding a perimeter and specifying a subset of foreground portions and a subset of background portions, the perimeter portion of the mask being user-defined; specifying a threshold value for the outer perimeter to separate voxels based on the voxel intensity, the threshold value being user-defined; filtering the medical image using the mask and the threshold to obtain the filtered image; and displaying the filtered image on a display.

Images