Method for electroscopic imaging for the analysis of cells - Patents.com

The method of applying a pH step change and image segmentation on ChemFET sensor arrays improves cell detection and analysis by accurately distinguishing cellular from non-cellular regions, enhancing data processing efficiency.

JP2024542936APending Publication Date: 2024-11-19LIFE TECHNOLOGIES CORP
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
JP2024521274
Authority / Receiving Office
JP · JP
Patent Type
Applications
Current Assignee / Owner
Priority Date
2021-10-27
Filing Date
2022-10-27
Publication Date
2024-11-19

Smart Images

  • Figure 2024542936000001_ABST
    Figure 2024542936000001_ABST
Patent Text Reader

Abstract

Analyzing cells disposed on a surface of a ChemFET sensor array may include flowing a solution having a step change in pH across the sensor array surface, where the ChemFET sensors of the sensor array generate signals in response to the step change in pH to generate electroscopic image data. During an acquisition time interval, multiple frames of electroscopic image data are acquired. Each frame corresponds to a signal sample generated by the sensor array measured at a sampling time during the acquisition time interval. Each frame includes pixels, with a given pixel in the frame corresponding to a signal sample from a given sensor in the sensor array. The electroscopic image data is segmented based on characteristics of the signal samples into cell regions corresponding to the location of cells on the sensor array surface and background regions corresponding to areas on the sensor array that do not have cells.
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] (CROSS REFERENCE TO RELATED APPLICATIONS) This application claims the benefit of U.S. Provisional Application No. 63 / 272,361 under 35 U.S.C. § 119(e), filed October 27, 2021. The entire contents of the aforementioned application are incorporated herein by reference. Summary of the Invention

[0002] A method for analyzing cells disposed on a sensor array surface of a ChemFET sensor array device may include flowing a solution having a step change in pH across the sensor array surface, where a plurality of ChemFET sensors of the ChemFET sensor array generate a plurality of signals in response to the step change in pH of the flowed solution to generate electroscopic image data; acquiring a plurality of frames of electroscopic image data during an acquisition time interval, where each frame corresponds to a signal sample of a plurality of signals generated by the ChemFET sensor array measured at a sampling time during the acquisition time interval, where each frame includes a plurality of pixels, where a given pixel in the frame corresponds to a signal sample from a given sensor in the ChemFET sensor array; and segmenting the electroscopic image data into one or more cell regions corresponding to the location of cells on the sensor array surface and one or more background regions corresponding to areas on the sensor array that do not have cells based on characteristics of the signal samples generated in response to the step change in pH of the flowed solution. The novel features are set forth with particularity in the appended claims. A better understanding of the features and advantages of the present teachings will be obtained by reference to the following detailed description that sets forth illustrative embodiments and the accompanying drawings. [Brief description of the drawings]

[0003] [Figure 1A]FIG. 1 is a block diagram depicting a chemical field effect transistor (ChemFET)-based cell analysis system for generating electroscopic images in accordance with various embodiments of the systems and methods of the present teachings. [Figure 1B] FIG. 1 is a schematic depiction of cells disposed on a cellular analysis system of the present teachings. [Figure 1C] 1 is a schematic cross-sectional view of a cell disposed on a sensor device of the present teachings. [Figure 2A] 1 illustrates an example of an output signal from a sensor array during an acquisition time interval represented as a series of 2D images. [Figure 2B] 1 is a table summarizing the attributes of an exemplary ChemFET sensor. [Figure 2C] 1 is a table displaying ChemFET system attributes for cell analysis across a range of cell sizes. [Diagram 3] FIG. 2 is a block diagram showing a general overview of processing electroscopic image data according to one embodiment. [Figure 4A] FIG. 3 is a block diagram of electroscopic image segmentation 304, according to one embodiment. [Figure 4B] FIG. 1 is a block diagram for forming a consensus cell mask, according to one embodiment. [Figure 4C] FIG. 13 is an exemplary block diagram for forming a final consensus cell mask. [Diagram 5] 1 shows an example (not to scale) of a convolution kernel that may be used to calculate the local spatial average of pixel values. [Figure 6A] 1 shows an example grayscale image of a tile responsive to a CellFind flow. [Figure 6B] 13 shows an example of a coarse cell mask for a tile. [Figure 6C] 13 shows an example of a tile of a local background corrected cell mask. [Figure 6D] 1 is an example of a sensor signal corresponding to a single pixel in response to a CellFind flow. [Figure 6E] 1 shows an example of a tile with cell regions and background regions. [Figure 6F]6E shows an example plot of the sensor signal corresponding to two pixels in the tile of FIG. 6E in response to a CellFind flow over time. [Figure 7] FIG. 1 is an illustrative diagram of an example of a single cell region and a local background region. [Figure 8] FIG. 2 is an illustrative diagram of an example of two cellular regions and two local background regions. [Figure 9] FIG. 1 is an exemplary block diagram for extracting signal data from a cellular region of interest. [Figure 10] 4 shows examples of tile images, segmentation results, and extracted signal data. [Figure 11A] 13 shows an example plot of the mean and standard deviation of averaged ROI signal values ​​over time. [Figure 11B] An example bar graph of maximum average ROI signal values ​​for multiple cells within the size range of 42-64 µm is shown. [Figure 12A] 1 shows an example of a spatial plot of maximum amplitude of pixels within a cell region over an acquisition period. [Figure 12B] 1 shows an example of a spatial plot of the minimum amplitude of pixels within a cell region over an acquisition period. [Figure 12C] 1 shows an example of a spatial plot of peak value versus time for pixels within a cell region over an acquisition period. [Figure 12D] 1 shows an example of a spatial plot of trough values ​​versus time for pixels within a cell region over an acquisition period. [Figure 13] An example of cell size histograms and statistics for 101 cells is shown. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0004] Electroscopic imaging of cells using a ChemFET sensor array-based system may include plating a sample of cells on a sensor array surface of a ChemFET sensor array device mounted in a flow cell. Each cell in the sample of cells has a footprint on the sensor array surface. The ChemFET sensor array-based system is configured to output a signal of each sensor in the ChemFET sensor array during an experiment on the sample of cells. The output signal of each sensor is instantaneously sampled by an analog-to-digital converter. The sampled signal for the ChemFET sensor array at a particular instant in time may be represented as a two-dimensional electroscopic image.

[0005] FIG. 1A generally illustrates a block diagram of exemplary components of a cellular analysis system 100 in accordance with the present teachings. As depicted in FIG. 1A, the cellular analysis system 100 can include various fluidic systems, as well as an array controller 50, a user interface 60, and a sensor array device assembly 115. As described in more detail herein, various cellular analyses can be performed using the sensor array device 110 of the sensor array device assembly 115. The structure and / or design of the cellular analysis system for use with the present teachings may include one or more features described in U.S. Patent Application Publication No. 2020 / 0088676, filed March 19, 2020, which is incorporated herein by reference in its entirety.

[0006] FIG. 1B generally illustrates a pictorial representation of a cellular analysis system 100A. As depicted in FIG. 1B, a cell 5 with a nucleus 2 and cell membrane 4 is disposed on a plurality of subsets of microwells, thereby defining a contact area or footprint that the cell 5 occupies on a corresponding subset of sensors. As recited herein, "contact area" and "footprint" can be used synonymously. The cellular analysis system 100A shares many of the same features as described for the schematic depiction of the cellular analysis system 100 of FIG. 1A. The cellular analysis system 100A of FIG. 1B can include a reagent fluid system 10 and a wash solution fluid system 20. The reagent fluid system 10 can include a plurality of reagent vessels, such as reagent vessels 11A-11D of FIG. 1B. Each reagent vessel can be in fluid communication with a reagent fluid line, such as reagent fluid lines 13A-13D of FIG. 1B. Flow from each reagent fluid line of the cell analysis system of the present teachings can be controlled by a valve, such as reagent fluid line valves 15A-15D of FIG. 1B. Wash solution fluid system 20 can include wash solution container 21, which can contain a wash solution of known electrolyte composition, as well as wash solution fluid line 23, wash solution fluid line valve 25, and reference electrode 27 in wash solution fluid line 23. As described in more detail herein, reference electrode 27 can provide a stable reference voltage 127 to sensors in the sensor array device. Thus, sensor array device 110 of FIG. 1B can be in fluid communication with reagent fluid system 10 and wash solution fluid system 20. Additional electrodes, not shown in FIG. 1B but which can be in communication with the sensor array device, can be utilized to provide electrical stimulation to cells on the sensor array, such as cell 5 of FIG. 1B.

[0007] The sensor array device 110 may include a sensor array or pixel array 120. As listed herein, the terms "sensor" and "pixel", as well as the terms "device" and "chip", and derivatives of these terms, may be used interchangeably. Additionally, the terms "sensor array" and "ChemFET sensor array", and derivatives thereof, may be used interchangeably. Although FIG. 1B depicts a two-dimensional array as a regular array, various embodiments of the sensor array of the present teachings may be arranged in various array shapes, for example, in a hexagonal close-packed shape. The sensor array device 110 may include a microwell array 122, which depicts each microwell that cooperatively engages with each sensor or pixel in the sensor array 120, as illustrated in FIG. 1B, such that each of microwells 122A1 to 122A6 cooperatively engages with a corresponding sensor 120A1 to 120A6. However, in various embodiments of the sensor array device of the present teachings, there may be more than one pixel per well. As described in more detail herein, various types of sensor array devices of the present teachings can be fabricated with defined but different microwell depths. Still other types of sensor array devices of the present teachings may not have microwell structures formed on the sensor array. Each sensor of the sensor array 120 can have a sensing surface in fluid communication with fluid in the microwell array 122. In various embodiments of the cellular analysis system 100 of the present teachings, each sensor of the sensor array 120 can be a chemical field-effect transistor (ChemFET), with each sensor in the sensor array 120 including at least one chemically sensitive field effect transistor. In accordance with the present teachings, the sensor array 120 can include a ChemFET that can be fabricated with a sensing surface selectively modified for the analysis of target chemical species of interest in cell biology, such as, for example, glucose, sucrose, lactate, and urea.As another non-limiting example, an ion sensitive field-effect transistor (ISFET) can have a sensing surface modified to be selective for various ions of interest, particularly for various cell metabolism studies such as hydrogen, potassium, calcium, and chloride.

[0008] In that regard, the inventors have recognized that various embodiments of the cell analysis system of the present teachings can be used to monitor changes in, for example, cellular electrophysiology and metabolism of cells subjected to any of a variety of conditions or stimuli. Moreover, the inventors have recognized that changes in the state of a cell that can cause a change in the electrical potential of the sensing surface of the ChemFET sensor can be monitored by the sensors of the various embodiments of the ChemFET sensor array of the present teachings. For example, the inventors have recognized that a cell is capacitively coupled to the sensing surface of the sensor such that when the electrical potential across the cell membrane changes in response to a chemical or electrical stimulus, the change in the electrical potential across the cell membrane can be detected by the sensors of the various embodiments of the ChemFET sensor array of the present teachings. Additionally, any changes, for example, changes in cell metabolism that can cause a change in the electrical potential of the sensing surface of the ChemFET sensor can be detected by the sensors of the various embodiments of the ChemFET sensor array of the present teachings. As described in more detail herein, such changes may be detected locally in relation to the contact area or footprint of a cell immobilized on the sensor array surface, such as in the case of cell efflux that may flow from a cell in response to a condition or stimulus, or may be detected in areas not associated with the cell footprint.

[0009] Data collected in experiments monitoring cellular responses to various stimuli with various embodiments of the ChemFET sensor array device of the present teachings can be presented to the end user in a number of formats. In one format, the time response is presented as detector counts, which can be easily correlated with millivolt (mV) changes in the sensing surface potential as a function of time. In another format, the spatial visualization of the cells can be presented as an electroscopic image for any selected time over the course of a selected application. The inventors have recognized that since electroscopic imaging is based on the various responses that can be induced in living cells, it can be useful as a general tool for visualizing cells on a sensor array, for example. For example, by reviewing the electroscopic image of cells fixed to the sensor array, the end user can select an area of ​​interest as part of the application configuration before running an experiment. As described in more detail herein, windowing down a selected area of ​​the sensor array device increases the data rate of the experiment. In accordance with the present teachings, substantial pixel coverage over the footprint of a cell coupled with high data rates can provide intracellular monitoring of action potentials of various excitable cells, for example, where data rates in the sub-millisecond range may be required.

[0010] 1C, a partial cross-sectional view of the sensor array 120 is depicted with a first sensor 120-1 and a second sensor 120-2. In various embodiments of a sensor array device of the present teachings, the sensor array 120 can include a floating gate top 130 coupled to a sensor floating gate structure 140. Alternatively, in various embodiments of a sensor array device of the present teachings, the sensor array 120 can include a sensor floating gate structure 140. As described in more detail herein, the floating gate top 130 can include a top metal layer, a sensor plate 132, and a metal via 134 formed in a dielectric 135.

[0011] The sensor floating gate structure 140 can have a metal layer 136 coupled to the sensor plate 132 through a metal via 134. The metal layer 136 is a top floating gate conductor in the sensor floating gate structure 140. In the illustrated example, the sensor floating gate structure 140 includes multiple layers of conductive material within a layer of dielectric material 150. The sensors 120-1 and 120-2 can include conductive terminals including source / drain regions 142 and 144 in a semiconductor substrate 160. The source / drain regions 142 and 144 include doped semiconductor material having a different type of conductivity than the conductivity type of the substrate 160. For example, the source / drain regions 142 and 144 can include doped P-type semiconductor material, and the substrate 160 can include doped N-type semiconductor material. A channel region 152 separates the source / drain regions 142 and 144. The floating gate structure 140 covers the channel region 146 and is separated from the substrate 160 by a gate dielectric 152. The gate dielectric 152 can be, for example, silicon dioxide. Alternatively, other suitable dielectrics can be used for the gate dielectric 152, such as materials having a higher dielectric constant, silicon carbide (SiC), silicon nitride (Si3N4), silicon oxynitride (Si2N2O), aluminum nitride (AlN), hafnium dioxide (HfO2), tin oxide (SnO2), cesium oxide (CeO2), titanium oxide (TiO2), tungsten oxide (WO3), aluminum oxide (Al2O3), lanthanum oxide (La2O3), gadolinium oxide (Gd2O3), and any combination thereof.

[0012] As described in more detail herein, the sensing surface 126S of the sensor plate 132 can function as a sensor surface for monitoring changes in, for example, cellular electrophysiology and metabolism of cells subjected to any of a variety of conditions or stimuli. In that regard, a cell 5, shown in FIG. 1C as a partial cross-section of a cell, is depicted as being disposed on the sensor plate 132 of the sensors 120-1 and 120-2. The cell 5 is depicted as being immobilized on the sensor array 120 via a surface coating 124. The surface coating 124 can be any cytocompatible material, such as various biopolymer materials including poly-D-lysine, laminin, fibronectin, collagen, and combinations thereof, as well as various preparations of extracellular matrix (ECM). An end user can perform applications using the cell analysis system of the present teachings, in which various reagents and solutions can be controllably flowed over the surface of the sensor array 120, as indicated by the arrows at the top of FIG. 1C.

[0013] The sensors 120-1 and 120-2 respond to changes in the surface potential of the ionic layer 128 proximate the sensing surface 126S, which can cause a change in the voltage of the floating gate 140. The applied reference voltage therefore ensures that the voltage of the floating gate exceeds a threshold voltage, providing that a small change in the floating gate voltage can cause a current to flow through the channel region 146, as described earlier in this specification with respect to FIG. 1B, resulting in an output signal of the sensors 120-1 and 120-2. In that regard, changes to the surface potential of the ionic layer 128 can be measured, for example, by measuring the current in the channel region 146 between the source region 142 and the drain region 144. The sensors 120-1 and 120-2 can therefore be used to directly provide a current-based output signal on an array line connected to the source region 142 or the drain region 144, or indirectly provide a voltage-based output signal with additional circuitry.

[0014] As described herein, changes in the state of the cell 5 that can alter the surface potential at the ionic layer 128 can be monitored by various embodiments of the ChemFET sensor array device of the present teachings. In terms of output signals, cellular activity that can increase the surface potential will result in a positive amplitude output signal of the ChemFET sensor, while cellular activity that can decrease the surface potential will result in a negative amplitude output signal of the ChemFET sensor. In that regard, changes in the state of the cell that can alter the surface potential of the ChemFET sensor can result in a measurable output signal. For example, metabolic activity that can increase the ion concentration of a cationic species for which the ISFET sensor is selective causes an increase in the surface potential. The result is a positive amplitude output signal of the ISFET sensor. Conversely, metabolic activity that can decrease the ion concentration of a cationic species for which the ISFET sensor is selective causes a decrease in the surface potential. The result is a negative amplitude output signal of the ISFET sensor. In another example, the surface potential can be altered by capacitive coupling of a cell 5 to the sensor array 120, so that when the potential across the cell membrane 4 changes in response to a chemical or electrical stimulus, the change in potential across the cell membrane can be detected by ChemFET sensors 120-1 and 120-2.

[0015] The output signal from the sensor array over the acquisition time interval may be represented as a series of two dimensional (2D) images, with each image corresponding to an array of signal samples received from the sensor array at a given time, similar to a movie. FIG. 2A illustrates an example of an output signal from the sensor array during the acquisition time interval as a series of 2D images 202. A 2D image 204 at a particular sampling time may be referred to herein as a frame, similar to a frame of a movie or video. Each 2D image is referred to herein as an electroscopic image. Each 2D image 204 may include multiple tiles 206 and multiple pixels 208. Since a frame is generated at each sampling time, the terms "sampling time" and "frame time" are used interchangeably herein. For example, 15 frames per second (fps) may be acquired during the acquisition time interval. For a 7 second acquisition interval, 105 frames are acquired at each acquisition time interval. Exemplary dimensions are 640 x 664 pixels for a tile, where 8 tiles x 12 tiles can contain the entire image for a frame. In some embodiments, the tiles can correspond to physical tiles of a ChemFET sensor array, where the physical tiles have dimensions of approximately 2 mm x 2 mm.

[0016] FIG. 2B shows Table 1 summarizing the attributes of an exemplary ChemFET sensor. Various embodiments of the sensor array device of the present teachings can have between about 20M and about 660M pixels, with a center-to-center spacing, or pitch, between each sensor that is between about 850 nm and about 3.36 μm. With respect to data collection, the collection of sensor output signals from all the sensors in the array constitutes a frame of data. For various sensor array devices of the present teachings, at about 20 million pixels to about 660 million pixels, a frame is a sizeable data file that is collected in units of hertz (Hz) as frames per second. Furthermore, because there is an inverse relationship between the area of ​​interest selected, representing the number of pixels, and the rate at which data can be collected, the frame rate can be increased by selecting a smaller subset of pixels to monitor, i.e., by windowing down the area of ​​the sensor array device from which data is collected. The effect of windowing down is demonstrated in Table 1 by comparing the value entered in the penultimate column, which is the maximum frame rate for collecting data from the entire device, with the value entered in the last column, which is the maximum frame rate for collecting data from a single row of the device. Thus, by windowing down to collect data from a single row, the frame rate is effectively increased.

[0017] Additionally, as provided in Table 1, the only difference between Device A and Device B is the total number of sensors per device, with twice as many sensors per Device B versus Device A. As shown in Table 1, the frame rate for Device B is half that of Device A, consistent with an inverse relationship between the number of pixels and the rate at which data can be collected. Therefore, one can select a device with a desired frame rate to fit the application.

[0018] FIG. 2C shows a table summarizing the attributes of an exemplary sensor array device for cell analysis. Sensor array device attributes that can be modified to provide various sensor array devices of the present teachings include pixel dimensions, as well as the rate at which data can be collected from the sensor array device. FIG. 2C provides an overview of five categories of cells by size relative to four exemplary sensor array devices of various sensor (pixel) dimensions, as given in Table 1 for Device B through Device E. The five categories of cells are identified descriptively, as well as by average diameter and average footprint.

[0019] Inspection of FIG. 2C reveals that the average diameter is 5 μm and the average area is 20 μm 2 For a tiny cell immobilized on the sensor array surface, the minimum contact area or footprint corresponds to about 1 row and about 2 pixels for the Chip 1 device, increasing to 6 rows and 32 pixels for the Chip 4 device. Similarly, for a tiny cell with an average diameter of 10 μm and an average area of ​​78 μm 2 For small cells immobilized on the sensor array surface, the minimum contact area or footprint corresponds to about 3 rows and about 8 pixels for the Chip 1 device, increasing to 12 rows and 126 pixels for the Chip 4 device. 2 For medium-sized cells immobilized on the sensor array surface, the minimum contact area or footprint corresponds to about 7 rows and about 50 pixels for the Chip 1 device, increasing to 29 rows and 792 pixels for the Chip 4 device. 2 A large cell immobilized on the sensor array surface can have a minimum contact area or footprint corresponding to about 15 rows and about 201 pixels for Chip 1, which increases to 59 rows and 3,168 pixels for the Chip 4 device. Finally, a large cell with an average diameter of 100 μm and an average area of ​​7,854 μm 2For extra-large cells immobilized on the sensor array surface, the minimum contact area or footprint corresponds to about 30 rows and about 803 pixels for the Chip 1 device, increasing to 118 rows and 12,668 pixels for the Chip 4 device. From inspection of Figure 2C, the trend is toward increasing pixel coverage with increasing cell size and decreasing pixel size.

[0020] In terms of pixels, the percent pixel coverage column is the percentage of the area of ​​the cell that a single pixel covers. For a very small cell immobilized on the sensor array surface, a single pixel corresponds to 50% coverage on the Chip 1 device, while a single pixel corresponds to 3% coverage on the Chip 4 device. Similarly, for a small cell immobilized on the sensor array surface, a single pixel corresponds to 12% coverage on the Chip 1 device, while a single pixel corresponds to 0.8% coverage on the Chip 4 device. For a medium sized cell immobilized on the sensor array surface, a single pixel corresponds to 2% coverage on the Chip 1 device, while a single pixel corresponds to 0.1% coverage on the Chip 4 device. A large cell immobilized on the sensor array surface may have a single pixel that corresponds to 0.5% coverage on the Chip 1 device, while a single pixel corresponds to 0.03% coverage on the Chip 4 device. Finally, for oversized cells immobilized on the sensor array surface, a single pixel corresponds to 0.1% coverage in the Chip 1 device, whereas a single pixel corresponds to 0.008% coverage in the Chip 4 device. From inspection of Figure 2C, the trend is toward decreasing percentage of cell coverage per pixel with increasing cell size and decreasing pixel size.

[0021] Considering what is presented in the table of FIG. 2C, the selection of pixel coverage of an exemplary sensor array device of the present teachings can be made for various average cell diameters. For example, for cells from about 5 μm to about 100 μm, the selection of a sensor array device can be made to provide coverage from about 8 pixels over 3 rows of pixels to 12,668 pixels over 118 rows of pixels for the corresponding footprint of the cell immobilized on the sensor array surface. Over that range of cell sizes, pixel sizes can vary, so that each pixel of the selected sensor array device can cover from about 12% of the cell to about 0.008% of the cell. Based on the data presented in FIG. 2C, it is clear that for any cell size, an exemplary sensor array device can be selected that can provide a significant number of sensors related to the contact area that the cell can occupy on the sensor array device. The spatial resolution that can be provided by the various sensor array devices of the present teachings can enable intracellular discrimination of signals, thus providing intracellular analysis.

[0022] With respect to data collection, in various cellular analysis systems of the present teachings, the collection of sensor output signals from all sensors in the array constitutes a frame of data. Given that various sensor array devices of the present teachings can have from about 20 million pixels to about 660 million pixels, a frame of data from various sensor array devices of the present teachings is a data file of significant size. Additionally, various cellular analysis systems of the present teachings include control circuitry coupled to the sensor array device that is configured to generate a significant number of frames of data from the sensor array device per second. Moreover, since there is an inverse relationship between the area of ​​interest selected and the rate at which data can be collected, the frame rate can be increased by selecting a smaller subset of pixels to monitor, i.e., by windowing down the area of ​​the sensor array device from which data is collected.

[0023] For example, referring to Table 1, a sensor array device with 40 million pixels can collect data at a frame rate of about 120 frames per second (fps). Then, when an area of ​​interest of 20 million pixels is selected, data at a frame rate of about 240 frames per second (fps) can be collected, and when an area of ​​interest of a single sensor array row is selected, data at a frame rate of about 75,000 frames per second (fps) can be collected. Specifically, for the exemplary sensor array device of the present teachings presented in FIG. 2C, the maximum frame rate per row of the sensor is provided in the penultimate column. In the last column, the maximum frame rate at which data can be collected for a fraction of the rows covered by cells, derived by dividing the maximum frame rate per row by the rows per cell diameter, is presented.

[0024] As can be seen by inspection of the last column of FIG. 2C, a significant number of frames per second can be collected for target areas of interest across a variety of cell sizes, providing data collection comfortably in the kHz range. For very small cells immobilized on the sensor array surface, data can be collected at a maximum frame rate of 75,000 fps on the Chip 1 device, whereas data can be collected at a maximum frame rate of 27,000 fps on Chip 4. Similarly, for small cells immobilized on the sensor array surface, data can be collected at a maximum frame rate of 25,000 fps on the Chip 1 device, whereas data can be collected at a maximum frame rate of 13,500 fps on Chip 4. For medium-sized cells immobilized on the sensor array surface, data can be collected at a maximum frame rate of 10,714 fps on the Chip 1 device, whereas data can be collected at a maximum frame rate of 5,587 fps on Chip 4. A large cell immobilized on the sensor array surface can have a maximum frame rate of 5,000 fps on the Chip 1 device, while data can be collected at a maximum frame rate of 2,746 fps on Chip 4. Finally, for an extra-large cell immobilized on the sensor array surface, data can be collected at a maximum frame rate of 2,500 fps on the Chip 1 device, while data can be collected at a maximum frame rate of 1,373 fps on Chip 4.

[0025] From inspection of FIG. 2C, the trend is for frame rate to decrease with increasing cell size and decreasing pixel size, which is consistent with an inverse relationship between the area of ​​interest selected and the rate at which data can be collected. Therefore, frame rate can be increased by selecting a smaller subset of pixels to monitor, i.e., by windowing down the area of ​​the sensor array device from which data is collected. Additionally, one can refer to Table 1 to select a device with a desired frame rate that matches the application.

[0026] FIG. 3 is a block diagram illustrating a general overview of processing electroscopic image data, according to one embodiment. The electroscopic image data includes a series of two-dimensional images 202 acquired from the sensor array over an acquisition time interval. In one example, the time interval may be seven seconds long. The acquisition time interval may be separated by a pause interval. The pause interval may allow data transfer of the electroscopic image data from the sensor array to the system processor. In some embodiments, the pause interval may be part of the experimental design. For example, the experiment may acquire high speed data at periodic intervals separated by a pause interval. For example, the experiment may acquire seven seconds of 120 fps data and have a pause interval of five minutes to sample high speed transient activity over a longer period of time. The data acquisition step 302 may acquire electroscopic image data including signal samples from individual sensors of the sensor array in parallel with each sampling time of the acquisition time interval. The data acquisition step 302 may acquire electroscopic image data over one or more acquisition time intervals to generate one or more sequences of electroscopic images. The one or more sequences of electroscopic images may be stored in memory for analysis by the processor.

[0027] The data acquisition step 302 may include receiving electroscopic image data from the sensor array in response to the "CellFind" flow. The CellFind flow may include a solution having a specific pH step difference from a buffer medium used to coat the sensor array surface on which the sample of cells is deposited. For example, the buffer medium may have a pH ranging from 4 to 8 units. For example, Thermo Fisher Live Cell Imaging Solution (Thermo Fisher Scientific, Cat. No. A14291DJ) may be used for the buffer medium. To achieve the specific pH difference for the CellFind flow, a small amount of NaOH or HCl is added to adjust the pH of the solution. The solution having the pH step difference is flowed across the sensor array. For example, the solution may have a pH difference of 0.01 to 0.5 units. The CellFind solution is adjusted relative to the medium and may be either acidic or basic relative to the buffer medium. For example, the pH change may be from 7.4 in the buffer medium to 7.3 in the CellFind flow solution. The pH change may be positive or negative. The ChemFET sensor generates a signal in response to the change in pH. In some embodiments, the polarity of the measured signal can be changed such that the response is in a positive direction even if the pH step change is in a negative direction. Areas of the sensor array that are not covered with cells respond more quickly to the pH step difference. Areas of the sensor array surface that are covered with one or more cells have a slower response to the pH step difference in the CellFind flow due to blockage of the sensor by the cells. Data acquisition during one or more acquisition time intervals that occur after the CellFind flow provides electroscopic image data over time that shows dynamic changes related to the location of the cells.

[0028] FIG. 6A shows an example of a single frame grayscale image of a tile's response to CellFind flow. FIG. 6D shows an example of a plot of a sensor signal corresponding to a single pixel in response to CellFind flow over time. FIG. 6E shows an example of a tile with a cell region and a background region. The arrows in FIG. 6E indicate that reagents, such as for the CellFind flow, flow from left to right. FIG. 6F shows an example of a plot of a sensor signal corresponding to two pixels in the tile of FIG. 6E in response to CellFind flow over time. The sensor signal corresponding to a pixel in the cell region 610 of FIG. 6E produced plot 614 of FIG. 6F. The sensor signal corresponding to a pixel in the background region 608 of FIG. 6E produced plot 612 of FIG. 6F. The time axis of FIG. 6F is in frames (15 fps). In this example, CellFind reagents flow for 2 seconds (from frame 15 to frame 45), which spans both the rising and falling sensor signals in response to the change in pH of the CellFind flow. The examples in Figures 6E and 6F show that areas of the sensor array covered by cells have a slower response to the CellFind flow than background areas not covered by cells.

[0029] The processor may apply an electroscopic image segmentation step 304 to the electroscopic image data to segment the 2D image into one or more cell regions corresponding to the location of cells on the sensor array and background regions corresponding to areas on the sensor array that do not have cells. A cell and background signal selection step 306 may select signals corresponding to the location of cell regions and background regions in the 2D image. In the following, the terms "image" and "tile" may be used synonymously since both are images. A signal parameterization step 308 may combine and parameterize signal samples corresponding to pixels in the cell regions. A user display and interaction step 310 may present the selected signals and parameterization to the user.

[0030] FIG. 4A is a block diagram of electroscopic image segmentation 304, according to one embodiment. In step 402, the gain may be corrected for each pixel in a frame. At the beginning of each experiment, the electrical gain of each pixel is determined by applying an external perturbation to the fluid potential and comparing the measured step size per pixel to the expected step size. The gain of each pixel may be corrected by dividing the pixel value in the frame by the electrical gain of the pixel. In step 404, a background value for each pixel is subtracted from each pixel of the frame. For example, a global background value per frame may be determined by calculating the statistical mode of all pixel values ​​in the frame. A global background value is calculated and subtracted from each pixel per frame for all frames acquired during the acquisition interval. In step 406, pixel values ​​of a 2D CellFind image (CI) may be calculated over the acquisition time interval. The pixels of the CellFind image include feature values ​​calculated based on characteristics of signal samples measured in response to the CellFind flow. The pixels of the CellFind image may be thresholded to form a binary CellFind image, where pixel values ​​equal to or greater than the threshold are assigned a value of 1 and pixel values ​​less than the threshold are assigned a value of 0. A pixel value of 1 in the CellFind image indicates the possible presence of a cell. A CellFind image may be determined for each tile for each frame within the acquisition time interval. In some embodiments, multiple CellFind images corresponding to the same tile may be calculated, with each CellFind image measuring a particular property, as described below.

[0031] A blurring function may be applied to the calculated CellFind image at step 408. In some embodiments, the CellFind image may be blurred by applying a low pass filter in the frequency domain as follows: A. For each tile of the CellFind image, a 2D Fourier transform, eg, a 2D Fast Fourier Transform (FFT), is computed to form a frequency domain image. B. Apply a low pass filter in the frequency domain to form a frequency domain filtered image. For example, a Butterworth filter may be applied. An exemplary Butterworth filter equation is given by: exp(-r 2.5 / 1.5N 3 ) (1) Here, r=X 2 +Y 2 where X and Y are coordinates in the frequency domain, and N is the minimum number of components in either X or Y. C. Compute the inverse 2D Fourier transform, e.g., the inverse 2D FFT, of the frequency domain filtered image. D. Retain the real part of the inverse 2D FFT to form the pixel values ​​of the blurred CellFind image.

[0032] In step 410, a threshold T may be applied to the pixel values ​​of the blurred CellFind image to generate a binary image having 1s at pixel locations where pixel values ​​in the blurred CellFind image are greater than or equal to T, and 0s at pixel locations where pixel values ​​in the blurred CellFind image are less than T. T = (mean + [max-algo_min] * frac) (2)

[0033] In equation (2), the mean and max of pixel values ​​of the blurred CellFind values ​​of the tile may be determined. The fraction "frac" may be defined by the user. For example, the frac value may be set to 0.1. The resulting binary CellFind image provides a coarse cell mask 412, where the approximate locations of cells are indicated by 1's and background regions are indicated by 0's.

[0034] In some embodiments, the CellFind image may include a "peak-to-peak (PTP)" image determined for frames within the acquisition time interval. The PTP value of a pixel in the PTP image may be determined by the following steps: a. For each pixel location in the tile during the acquisition time interval, subtract the minimum pixel value from the maximum pixel value. b. Blur the PTP image (step 408). c. Apply a threshold T to the pixel values ​​of the blurred PTP image according to equation (2) (step 410). For example, algo_min may be set to 100, which is 25% of a typical maximum PTP value. algo_min may range from 10% to 100% of the maximum value of the PTP image. d. Pixel values ​​greater than or equal to the threshold are assigned a value of 1 and pixel values ​​less than the threshold are assigned a value of 0 to form a binary CellFind image (step 412).

[0035] In some embodiments, the CellFind image includes a "Peak-to-Peak Absolute Value (PTP-Ab)" image determined for frames within the acquisition time interval. The PTP-Ab image may be determined by the following steps: a. Calculate the PTP images for the acquisition time interval. b. Calculate the average PTP value of the pixels in the PTP image. c. The average PTP value is subtracted from the PTP value of each pixel in the PTP image to form a difference image. d. Calculate the absolute value of the difference for each pixel in the difference image to form an absolute difference image. e. The average PTP value is added to each pixel to form a PTP-Ab image. f. Blur the PTP-Ab image (step 408). g. Apply a threshold T to the pixel values ​​of the blurred PTP-Ab image according to equation (2) (step 410). For example, algo_min may be set to 100. algo_min may range from 10% to 100% of the maximum value of the PTP-Ab image. An algo_min value of 100 is typically 25% of the maximum PTP-Ab value. Pixel values ​​that are greater than or equal to the threshold are assigned a value of 1 and pixel values ​​that are less than the threshold are assigned a value of 0 to form a binary CellFind image (step 412).

[0036] In some embodiments, the CellFind image may include a "maximum variation (MaxVar)" image determined for the frames within the acquisition time interval. The MaxVar image may be determined by the following steps: a. Calculate the average value of the pixels in the tile at each frame or sampling time to give an array of average values. For example, for 105 frames within the acquisition time interval, 105 average values ​​are calculated to give a 1×105 array of average values ​​corresponding to the tile. b. Calculate the differences between adjacent mean values ​​in the array of mean values ​​to form an array of difference values. For example, for a 1×105 array of mean values, a 1×104 array of difference values ​​is calculated. c. The frame with the largest difference in the array of difference values ​​is frame f. max Identify as. d. Tile (f max ) and the tile (f1) of the first frame in the acquisition time interval tile (f1). max )-tile(f1). e. Blur the MaxVar image (step 408). f. Apply a threshold T to the pixel values ​​of the blurred MaxVar image according to equation (2) (step 410). For example, algo_min may be set to 100. algo_min may range from 10% to 100% of the maximum value of the MaxVar image. Pixel values ​​that are greater than or equal to the threshold are assigned a value of 1 and pixel values ​​that are less than the threshold are assigned a value of 0 to form a binary CellFind image (step 412).

[0037] In some embodiments, the CellFind image may include a composite of MaxVar sub-images determined for subtiles of frames within the acquisition time interval, referred to herein as a "MaxVar local" image. The MaxVar local image may be determined by the following steps: Divide the tile into multiple subtiles. The subtile dimensions can be 5%-50% of the total tile dimensions. For example, for a tile dimension of 664x640, the subtile dimensions can be approximately 80x80. b. Compute the MaxVar local image for each subtile, independent of the other subtiles. An array of mean values ​​and an array of difference values ​​are computed for each subtile. max and subtile (f max ) is identified for each subtile. Each MaxVar local image is expressed as MaxVar local image = subtile (f max )-corresponding to subtile (f1). c. Generate a composite MaxVar local image from the determined MaxVar local images for the subtiles. d. Blur the composite MaxVar local image (step 408). e. Apply a threshold T to pixel values ​​of the blurred composite MaxVar local image according to equation (2) (step 410). For example, algo_min may be set to 100. algo_min may range from 10% to 100% of the maximum value of the composite MaxVar local image. f. Pixel values ​​greater than or equal to the threshold are assigned a value of 1 and pixel values ​​less than the threshold are assigned a value of 0 to form a binary CellFind image (step 412).

[0038] In some embodiments, the CellFind image may include a function of a time average of the pixels of the tile over the acquisition time interval to form a time-averaged image. The time-averaged image may be determined by the following steps: a. Calculate the average over the acquisition time interval of each pixel value in the tile to form a first average image. b. Calculate the average of each pixel value in the tile over multiple initial frames within the acquisition time interval to form a second average image. The number of frames in the multiple initial frames can be in the range of 5% to 50% of the total number of frames within the acquisition time interval. For example, the number of initial frames can be the first 10 frames and the total number of frames can be 105 frames within the acquisition time interval. c. The second average image is subtracted from the first average image to form a time-averaged image. d. Blur the time-averaged image (step 408). e. Apply a threshold T to the pixel values ​​of the blurred time-averaged image according to equation (2) (step 410). For example, algo_min may be set to 100. algo_min may range from 10% to 100% of the maximum value of the time-averaged image. f. Pixel values ​​greater than or equal to the threshold are assigned a value of 1 and pixel values ​​less than the threshold are assigned a value of 0 to form a binary CellFind image (step 412).

[0039] In some embodiments, the CellFind image may include a "time to peak (tPeak)" image. The tPeak image may be determined by the following steps: For each pixel location, determine the maximum pixel value during the acquisition time interval. b. For each maximum pixel value, determine the frame index or frame number that corresponds to the maximum pixel value. c. Form a tPeak image where each pixel location has the frame index or frame number corresponding to the maximum pixel value. d. Blur the tPeak image (step 408). d. Apply a threshold T to the pixel values ​​of the blurred tPeak image according to equation (2) (step 410). For example, algo_min may be set to 4 frames. algo_min may range from 1 to the total number of frames in the acquisition time interval. e. Pixel values ​​greater than or equal to the threshold are assigned a value of 1 and pixel values ​​less than the threshold are assigned a value of 0 to form a binary CellFind image (step 412).

[0040] In some embodiments, the CellFind image may include a "time to peak 80% (tPeak80)" image. The tPeak80 image may be determined by the following steps: For each pixel location, determine the maximum pixel value during the acquisition time interval. b. For each pixel location, calculate 80% of the maximum pixel value. c. Determine the frame index or frame number that corresponds to the 80% maximum pixel value. d. Form a tPeak80 image where each pixel location has the frame index or frame number corresponding to the maximum pixel value. Blur the e.tPeak80 image (step 408). f. Apply a threshold T to the pixel values ​​of the blurred tPeak80 image according to equation (2) (step 410). For example, algo_min may be set to 3 frames. algo_min may range from 1 to the total number of frames in the acquisition time interval. g. Pixel values ​​greater than or equal to the threshold are assigned a value of 1 and pixel values ​​less than the threshold are assigned a value of 0 to form a binary CellFind image (step 412).

[0041] In some embodiments, the CellFind image may include a "time to peak (tPeak local)" image. The tPeak local image may be determined by the following steps: For each pixel location, determine the maximum pixel value during the acquisition time interval. b. For each maximum value, determine the frame index or frame number that corresponds to the maximum pixel value to generate the tPeak image. c. Divide the tPeak image into subtiles. Subtile dimensions can be 5%-50% of the total tile dimensions. For example, for a tile dimension of 664x640, the subtile dimensions can be approximately 80x80. d. For each subtile, subtract the average pixel value within the subtile. e. Generate a composite tPeak local image containing the subtile. h. Blur the composite tPeak local image (step 408). i. Apply a threshold to the pixel values ​​of the blurred tPeak local image according to equation (2) (step 410). For example, algo_min may be set to 7 frames. algo_min may range from 1 to the total number of frames in the acquisition time interval. j. Pixel values ​​greater than or equal to the threshold are assigned a value of 1 and pixel values ​​less than the threshold are assigned a value of 0 to form a binary CellFind image (step 412).

[0042] In some embodiments, the CellFind image may include a "time to fall (tFall local)" image, which is an indication of the return to the original pH on the falling side of the signal. The tFall local image may be determined by the following steps: a. Divide the tile into subtiles. The subtile dimensions can be 5%-50% of the total tile dimensions. For example, for a tile dimension of 664x640, the subtile dimensions can be approximately 80x80. b. Calculate the average value of the pixels of the subtile at each frame or sampling time to give an array of average values. For example, for 105 frames within the acquisition time interval, 105 average values ​​are calculated to give a 1×105 array of average values ​​corresponding to the subtile. c. The frame index f corresponding to the maximum mean value in the array of mean values ​​corresponding to the subtile max Or find the time index. d. Subtile corresponding to frame index, subtile(f max ) e. Last frame f last The corresponding subtile in,subtile(f last ) for difference image = subtile(f max )-Subtile(f last ) to calculate f. Calculate the fraction of each pixel value of the difference image to give a fractional value. For example, the fractional value may be 25%. The value range of the fractional value is 5% to 90%. g. Subtile the fractional value (f last ) to form the threshold image of the subtile. h. For each pixel location in the subtile, when the pixel value is below the threshold value corresponding to the pixel location in the threshold image, the frame index f T Or the number of frames is determined. i. For each pixel location in the subtile, T From f max to form the tFall local image corresponding to the subtile. j. Combine the tFall local images corresponding to the subtiles to form a combined tFall image corresponding to the tile. k. Blur the composite tFall image (step 408). l. Apply a threshold to pixel values ​​of the blurred composite tFall image according to equation (2) (step 410). For example, algo_min may be set to 6 frames. algo_min may range from 1 to the total number of frames in the acquisition time interval. m. Pixel values ​​greater than or equal to the threshold are assigned a value of 1 and pixel values ​​less than the threshold are assigned a value of 0 to form a binary CellFind image (step 412).

[0043] In some embodiments, the Pearson Difference (PD) may be calculated from a binary CellFind image derived from any of the CellFind methods described above. An estimated binary mask divides pixels into initial estimates of "cell" and "background" regions. The Pearson Difference may be calculated for each pixel of the estimated CellFind image as follows: PD=P o,c -P o,b (3) Here, P o,c is the Pearson correlation between the object time series and the average time series of all pixels identified as cells in the binary mask, P o,b is the Pearson correlation between the object time series and the average time series of all pixels identified as background. In this case, the objects are the individual pixels. The Pearson correlation may be calculated as follows: P x,y =cov(x,y) / [std(x) * std(y)] (4) Here, P x,yis the Pearson correlation coefficient of time series x and time series y, cov(x,y) is the covariance of time series x and time series y, std(x) is the standard deviation of time series x, and std(y) is the standard deviation of time series y. A time series is calculated for the average of all pixels in the region defined as cells (typically 1×105 for 105 frames in the acquisition time interval). A time series is calculated for the average of all pixels in the region defined as background (typically 1×105 for 105 frames in the acquisition time interval). A time series is calculated for the object. In this case, an object can be a single pixel or an average of one or more pixels (typically 1×105 for 105 frames in the acquisition time interval). The Pearson Difference (PD) CellFind image calculated by equation (3) can be segmented into cell and background regions using Otsu's method (www.en.wikipedia.org / wiki / Otsu%27s_method). A histogram of the PD CellFind image can be calculated. This histogram may have two peaks corresponding to two classes, one for cellular regions and one for background regions. Otsu's method determines a single threshold that optimally separates pixels into two classes corresponding to the two peaks in the histogram of the PD CellFind image by minimizing the within-class variance. The threshold may be applied to the PD CellFind image to generate a binary PD CellFind image.

[0044] The i-th coarse cell mask from step 412 may be provided in the (i+1)-th iteration 414 for calculation of the local background value in step 404. The local background value may be determined by applying a convolution kernel to the pixel values ​​of the tile. In some embodiments, the convolution kernel may be a rectangular array of coefficients. For example, the convolution kernel may have a central coefficient of 0 and all other coefficients equal to 1. Convolving the kernel with the pixel values ​​of the tile and dividing by the number of 1's in the kernel generates a local spatial average of the background value at the pixel location corresponding to the central coefficient. FIG. 5 shows an example (not to scale) of a convolution kernel that may be used to calculate the local spatial average of the pixel values. The dimensions of the convolution kernel may be configurable by the user. For example, for a tile dimension of 640×664 pixels, the convolution kernel may have dimensions of 41 pixels in the horizontal dimension by 101 pixels in the vertical dimension. The horizontal dimension is parallel to the direction of the CellFind flow. The vertical dimension is perpendicular to the direction of the CellFind flow. CellFind flow proceeds from one side of the sensor array to the other. FIG. 6E shows an example of the direction of CellFind flow across a tile from left to right. In some embodiments, the number of pixels in the convolution kernel may be 6% of the number of pixels in the tile. In some embodiments, the number of pixels in the convolution kernel may range from 3% to 20% of the number of pixels in the tile. In some embodiments, pixel locations corresponding to a coarse cell mask value of 1 may be excluded from the convolution input and local background may be estimated in the absence of cells identified in the coarse cell mask. An array of local background values ​​may be subtracted from the pixel values ​​of the tile at the corresponding pixel locations to form a local background corrected tile. An array of local background values ​​may be determined for all tiles in each frame and subtracted from each frame of the acquisition time interval using the same coarse cell mask to generate the corresponding local background corrected frame.

[0045] In the (i+1)th iteration of step 406, a CellFind image value is calculated for each pixel in the local background corrected frame to form a local background corrected CellFind image, as described above.

[0046] In the (i + 1)-th iteration of step 408, the blurring function can be applied to the locally background-corrected CellFind image. In the (i + 1)-th iteration of step 410, the threshold is applied to the pixel values of the blurred locally background-corrected CellFind image such that a binary image having a value of 1 at pixel locations where the blurred locally background-corrected CellFind value is greater than or equal to T i+1 and a value of 0 at pixel locations where the blurred locally background-corrected CellFind value is less than T i+1 can be generated. In some embodiments, the threshold T i+1 can be calculated by Equation (5). T i+1 = (mean i+1 + [max i+1 − algo_min] * frac i+1 ) (5)

[0047] In Equation (5), for the pixel values of the blurred locally background-corrected CellFind values of the tile, the mean value (mean i+1 ) and the maximum value (max i+1 ) can be determined. The fraction “frac i+1 ” can be defined by the user. For example, the frac i+1 value can be set to 0.1. The range of the frac i+1 value is 0 < frac i+1 < 1. The resulting binary image provides a locally background-corrected cell mask, where the approximate locations of the cells are indicated by 1s and the background regions are indicated by 0s. In some embodiments, the locally background-corrected cell mask can be determined for the acquisition time interval.

[0048] FIG. 6A shows an example of a grayscale image of a tile responsive to the CellFind flow. FIG. 6B shows an example of a tile of a coarse cell mask. FIG. 6B represents a first (i=1) coarse cell mask 412 before local background correction. FIG. 6C shows an example of a tile of a local background corrected cell mask after two iterations (i+1=2). FIG. 6C represents a local background corrected cell mask after local background subtraction of step 404, a CellFind calculation of step 406 using the peak-to-peak (PTP) calculation described above, a blurring operation of step 408, and thresholding of step 410. The features represent approximate possible cell locations. The features in the local background corrected cell mask of FIG. 6C are more numerous and more emphasized than the features in the coarse cell mask of FIG. 6B. For example, the cluster of features 602 in FIG. 6B and the larger cluster of features 604 in FIG. 6C correspond to the cluster of features 601 in the grayscale image of the tile response.

[0049] In some embodiments, multiple cell masks may be generated by computing two or more of the types of CellFind images described above. Figure 4B is a block diagram for forming a consensus cell mask, according to one embodiment. In step 416, multiple different types of CellFind images may be computed to generate multiple cell masks, as described above with respect to steps 404-412 of Figure 4A. In step 418, the cell masks may be weighted and summed as follows: Weighted Sum Image = Σ j w j × Cell Mask j (6)

[0050] Cellular mask j The conversion from binary data type to floating-point data type in jIn step 420, a first threshold may be applied to the weighted sum image. To form a binary consensus cell mask, pixel values ​​equal to or greater than the first threshold are assigned a value of 1, and pixel values ​​less than the threshold are assigned a value of 0. For example, the first threshold applied to the weighted sum image may be set to a relatively large value of 1.8 to identify regions that are likely to be cells. In some embodiments, a second consensus cell mask may be generated in step 420 by applying a second threshold to the weighted sum image. For example, the second threshold may have a lower value than the first threshold, such as 0.7. Because the second threshold is lower, the second consensus cell mask allows for more possible cell regions while allowing for some error. The range of values ​​for the first and second thresholds may be min(w j )~Σ j w j For example, the first threshold is selected to have a low false positive rate for identifying the cells, and the second threshold is selected to have a low false negative rate for identifying the cells.

[0051] Table A shows the weights w j We give example values ​​for the corresponding weights w when no specific CellFind image type is used. j can be set to 0. The weight w j The value of can be selected by the user.

[0052] [Table 1]

[0053] FIG. 4C is an exemplary block diagram for forming a final consensus cell mask. In an embodiment where only one type of CellFind image is calculated, as described above with respect to FIG. 4A, the coarse cell mask is the input cell mask to the size selection step 422. In an embodiment where multiple types of CellFind images are calculated, as described above with respect to FIG. 4B, the consensus cell mask is the input cell mask to the size selection step 422. In the size selection step 422, the size of the features in the input cell mask that correspond to the size of the object of interest is found. For example, the size of the object of interest for the analysis of cells may range from 20 to 75 microns. The size of the object of interest is determined by the effective diameter d eff It can correspond to.

[0054]

number

[0055] In step 424, the filtered cell mask is segmented using a watershed algorithm. The watershed algorithm segments selected features into one or more cell regions. In the filtered cell mask, each selected feature has a pixel value of 1 and background regions have a pixel value of 0. The filtered cell mask is segmented as follows: a. For pixels of the filtered feature region in the filtered cell mask (i.e., pixels with pixel value 1), calculate the distance to the nearest background pixel (i.e., pixels with pixel value 0). b. Identifying one or more maxima of the calculated distances within a given filtered feature region, each maxima corresponding to a centroid. c. Classify each pixel in the filtered feature region based on its distance to one or more centroids of the filtered feature region. Assign each pixel in the filtered feature region to the closest centroid. Pixels that are assigned to the same centroid are classified as cell regions. d. Create a labeled mask image by assigning an index to pixel locations within each cell region determined in step c. For example, a first cell region may have a 1 assigned to its pixel locations, a second cell region may have a 2 assigned to its pixel locations, etc.

[0056] In step 426, the cell mask may be verified to generate a final cell mask. The cell mask resulting from segmentation step 424 may show an excess of cell regions. Some of the identified cell regions may be errors and actually correspond to background regions. In step 426, the cell mask generated by step 424 may be spatially divided into multiple sub-regions. The CellFind images acquired during the acquisition time interval may be spatially divided into corresponding multiple sub-regions. For example, the cell mask and the CellFind images may each be divided into four quadrants. The sub-regions may range from 1 to 100 sub-regions. For each cell region, the Pearson difference (Equation 3) may be calculated between the cell mean time series and the cell and background time series for the nearest sub-region. In this case, the "object" corresponds to a particular cell, and P o,c Or P o,b The variable “o” in is the average time series of pixels defined as belonging to a single particular cell. o,c The variable “c” in is the average time series of all pixels defined as belonging to any cell according to the first threshold applied to the weighted sum image in step 420. o,bThe variable "b" in is the average time series of all pixels defined as belonging to the background according to the second threshold applied to the weighted sum image in step 420. The resulting PD values ​​for each cell are stored in a 1×N array, where N is the total number of cells identified in the cell mask. In some embodiments, to add numerical stability, one or more "fake cells" may be generated by randomly selecting pixels from the background region. The size of the fake cells may range from 10 pixels to 100% of the background region. A typical size is 20% of the background region. The number of fake cells may range from 5% of the identified cell area in the cell mask to 100% of the identified cell area. A typical number is 10% of the identified cell area. For each fake cell, the Pearson difference may be calculated and added to the value from the real cell to generate a 1×(N+F) array, where F is the total number of fake cells added. The array of Pearson differences can be divided into two clusters, for example according to k-means clustering (www.scikit-learn.org / stable / modules / generated / sklearn.cluster.KMeans.html). The cluster with the larger average Pearson difference is determined to contain "real" cell regions, and the cluster with the smaller average value is determined to contain segmentation artifacts. If more than 30% of the "fake cells" are found in the "real cell" cluster, the real cell cluster is determined to contain segmentation artifacts, and the verified cell is not associated with that cluster.

[0057] The final binary cell mask can be used to select a set of pixels near the centroid of each of the cell regions for analysis of the signal data acquired during the acquisition period. The selected subset of pixels near the centroid is referred to herein as a cell region of interest (ROI). A cell ROI can be selected by selecting a number of pixels that surround the centroid and are contained within the segmented region of the cell. The number of pixels selected can be defined by the user. For example, the number of pixels can be 25. The number of pixels selected can range from 1 pixel to the number of pixels in the entire area of ​​the cell that has a label corresponding to the labeled mask image.

[0058] A local background region for each cell region may be identified. The local background region for each cell region may be assigned a background label in the labeled mask image generated in step d above. The signal data acquired for the background region during the acquisition period may be used to remove an offset from the signal data of the cell ROI. The local background region for each cell region in the binary cell mask may be determined as follows:

[0059] An outer boundary of a local background region for a given cell region may be determined. For the cell region of a given cell, a first multiple iterations of binary dilation of the cell region are performed. The first number of iterations may be selected by a user. For example, the number of iterations may be 25. The first number of iterations may range from 15 to 150 iterations. The resulting region defines an outer boundary of the local background surrounding the cell region. The first multiple iterations of binary dilation generate a first binary mask having ones inside the outer boundary and zeros outside the outer boundary.

[0060] An inner boundary of a local background region for a given cell region may be determined. A second multiple iterations of binary dilation is performed for the cell region of the given cell. The second number of iterations may be less than the number of iterations used to define the outer boundary and may be selected by the user. For example, the number of iterations may be 10. The number of iterations may range from 5 to 50 iterations. The resulting region defines an inner boundary of the local background surrounding the cell region. The second multiple iterations of binary dilation generate a second binary mask having ones inside the inner boundary and zeros outside the inner boundary.

[0061] Computing the binary difference between the second binary mask and the first binary mask may give a local background binary mask that partially or entirely surrounds a given cell region. The local background binary mask contains ones in the region between the inner and outer boundaries. The pixels of the local background binary mask may be annotated in a labeled mask image.

[0062] Any pixel defined as the local background for a given cell region is located at a minimum distance from any cell region in the cell mask. The minimum distance may be set by the user. For example, the minimum distance may be 25 pixels. A given cell region may not be defined as the background of any other cell region. The inner boundary of the local background for a given cell region is spatially separated from the boundary of the cell region such that the cell region and the local background are not adjacent to each other.

[0063] 7 is an illustrative diagram of an example of a single cell region and a local background region. A cell region 702 may be represented by a pixel value of 1 in a cell mask. An outer boundary 706 may be determined by a first binary dilation. An inner boundary 704 may be determined by a second binary dilation. In this example, a local background region 708 surrounds the cell region 702 and is spatially separated from the boundary of the cell region 702. FIG. 7 is for illustrative purposes only and is not to scale.

[0064] FIG. 8 is an illustrative diagram of an example of two cell regions and two local background regions. Cell regions 801 and 802 may be represented by pixel value 1 in the cell mask. An outer boundary 806 may be determined by a first binary dilation. An inner boundary 804 may be determined by a second binary dilation. In this example, a local background region 810 partially surrounds cell region 801 and a local background region 808 partially surrounds cell region 802. The local background regions 808 and 810 are spatially separated from the boundaries of cell regions 801 and 802. FIG. 8 is for illustrative purposes only and is not to scale.

[0065] In some embodiments, the second consensus cell mask may be generated by using a lower threshold in step 420, as described above with respect to FIG. 4B. The inverse of the second consensus cell mask may be calculated. A binary dilation may be applied to the inverse of the second consensus cell mask. The resulting inverse mask may be ANDed with the local background binary mask to form a second local background binary mask.

[0066] FIG. 9 is an exemplary block diagram for extracting signal data from a cellular region of interest. FIG. 10 shows an example of tile images, segmentation results, and extracted signal data. Image 1010 shows the final binary mask for a particular tile. Several cellular regions are shown, such as area 1011. In this example, area 1012 is selected for analysis of the signal data. Image 1020 is a close-up of area 1012 of the final cellular mask 1010. Two cellular regions are shown, namely a set of pixels of cellular region 1021 and a set of pixels of cellular region 1022. This example focuses on cellular region 1021. In step 902, a centroid of the set of pixels of cellular region 1021 may be calculated. The centroid calculation may exclude pixels at the boundary of cellular region 1021. The centroids are marked with an "X" in images 1020, 1030, and 1040. In step 904, a subset of pixels within the cell region 1021 may be selected to define a region of interest (ROI) 1031. For example, the subset of pixels selected for the ROI may surround the centroid. The number of pixels selected for the ROI 1031 may be selected by a user. For example, the selected number may be 25 pixels. In another example, a subset of pixels within the cell region 1021 may be selected by a user for the ROI.

[0067] In step 906, a signal corresponding to the ROI pixel may be obtained. The signal corresponding to the ROI pixel may include a sensor signal for a sensor at a corresponding location of the ChemFET sensor array. The signal may be measured by the sensor after a reagent is flowed over the sensor array. The signal measured by the sensor corresponding to the location of the cell may indicate the response of the cell to the reagent. For each pixel, there is one signal sample per frame during the acquisition time interval. In step 908, the values ​​of the signal samples may be spatially averaged for pixels in the ROI at each frame time or sampling time to provide an average signal value per frame. The number of pixels in the ROI may be calculated for each frame time t FRAMEDetermine the number of signal samples to be spatially averaged in . The spatial average may be calculated as follows: Avg_ROI_sig(t FRAME )=Σ ROI sig(t FRAME ,x ROI ,y ROI ) / N ROI (8) where Avg_ROI_sig(t FRAME ) at time t FRAME Average ROI signal value per frame in Σ ROI is the sum over pixels within the ROI (e.g., ROI1031), sig(t FRAME ,x ROI ,y ROI ) is the pixel coordinate (x ROI ,y ROI ) at time t FRAME The value of the signal sample at N ROI is the number of pixels in the ROI.

[0068] In step 910, a set of pixels in a local background region for a particular cell region may be identified in a local background mask. For example, image 1040 shows a local background region 1041 of cell region 1021. In step 912, signals corresponding to the background pixels may be obtained. The signals corresponding to the background pixels include a sensor signal for a sensor at a corresponding location of the ChemFET sensor array where no cells are present. A signal measured by the sensor after a reagent is flowed over the sensor array may represent the effect of the flowed reagent on the sensor. In step 914, the values ​​of the signal samples may be spatially averaged for pixels in the background region at each frame time or sampling time to provide an average signal value per frame. The number of pixels in the background region may be determined for each frame time t FRAME Determine the number of signal samples to be spatially averaged in . The spatial average may be calculated as follows: Avg_bkg_sig(t FRAME )=Σ BKG bkg_sig(t FRAME ,x BKG ,y BKG) / N BKG (9) where Avg_bkg_sig(t FRAME ) at time t FRAME The average background signal value per frame in Σ BKG is the sum over pixels in a local background region (e.g., local background region 1041), bkg_sig(t FRAME ,x BKG ,y BKG ) is the pixel coordinate (x BKG ,y BKG ) at time t FRAME The value of the background signal sample in N BKG is the number of pixels in the local background region.

[0069] In step 916, the average background signal value per frame is subtracted from the average ROI signal value per frame to obtain the cell average ROI signal per frame Cell_ROI_sig(t FRAME ) is given as follows: Cell_avg_ROI_sig(t FRAME )=Avg_ROI_sig(t FRAME )-Avg_bkg_sig(t FRAME ) (10)

[0070] Referring to FIG. 10, graph 1050 shows an example of a plot of the average ROI signal and the average background signal. Plots 1051, 1052, 1053, and 1054 show the average ROI signal values ​​for each data acquisition time interval. Plots 1061, 1062, 1063, and 1064 show the average background signal values ​​for each data acquisition time interval. Plots 1051, 1061, 1052, 1062, 1053, and 1063 show the average ROI signal values ​​and the average background signal values ​​in response to three flows of a washing solution over the sensor array. Plots 1054 and 1064 show the average ROI signal values ​​and the average background signal values ​​determined for HepG2 cells in response to flows of a reagent solution having increasing concentrations of FCCP over the sensor array. FCCP (carbonyl cyanide p-(trifluoromethoxy)phenylhydrazone) is a potent deactivator of oxidative phosphorylation in mitochondria. The concentrations of FCCP in the reagent solution (Thermo Fisher Live Cell Imaging Solution, Thermo Fisher Scientific, Cat. No. A14291DJ) were 1, 10, and 100 mM.

[0071] Returning to FIG. 3, in step 308, the average ROI signal value of the cells may be parameterized to analyze the amplitude and time response. For example, statistical parameters such as the mean, median, and standard deviation of the average ROI signal value of the cells over time may be calculated. FIG. 11A shows an example of a plot of the mean and standard deviation over time. The horizontal axis represents the time axis in seconds. Traces corresponding to acquisition periods, each having a duration of 7 seconds, are plotted. The time intervals between the acquisition periods have been removed. Trace 1102 plots the + / - standard deviation and trace 104 plots the mean. The mean and standard deviation are plotted for the response to four flows of the control "CON" solution and one flow of the glucose "Glu" solution. The drops in the average traces are transients as the solutions flow past the cells. The average trace in response to the glucose solution maintains an elevated steady state 1106 compared to the response to the control solution flow. Figure 11B shows an example bar graph of maximum average ROI signal values ​​for multiple cells in the size range of 42-64 µm. The results shown in Figure 11A are a subset of the results shown in Figure 11B.

[0072] 12A-12D show examples of pixel-by-pixel spatial plots of signal parameters for two cells. In these plots, the average local background is subtracted from the pixels in the cell region as shown in the final cell mask. The centroid of each cell is marked with an "X". FIG. 12A shows an example of a spatial plot of the maximum amplitude of pixels in the cell region over the acquisition period. FIG. 12B shows an example of a spatial plot of the minimum amplitude of pixels in the cell region over the acquisition period. FIG. 12C shows an example of a spatial plot of peak value of pixels in the cell region over the acquisition period versus time. FIG. 12D shows an example of a spatial plot of trough value of pixels in the cell region over the acquisition period versus time.

[0073] The final cell mask may be analyzed for cell size statistics. Figure 13 shows an example of cell size histograms and statistics for 101 cells. The abbreviations q1, q2, and q3 refer to the first, second, and third quartiles, respectively. The abbreviation "iqr" refers to the interquartile range, q3-q1, or the central half of the distribution. The effective diameter is described above with respect to equation (7).

[0074] According to various exemplary embodiments, one or more features of any one or more of the teachings and / or exemplary embodiments discussed above may be performed or implemented using appropriately configured and / or programmed hardware and / or software elements. The determination of whether an embodiment is implemented using hardware and / or software elements may be based on any number of factors, such as desired computation speed, power levels, thermal tolerance, processing cycle budget, input data rates, output data rates, memory resources, data bus speeds, etc., as well as other design or performance constraints.

[0075] Examples of hardware elements may include processors, microprocessors, input and / or output (I / O) devices (or peripherals) communicatively coupled via local interface circuits, circuit elements (e.g., transistors, resistors, capacitors, inductors, etc.), integrated circuits, application specific integrated circuits (ASICs), programmable logic devices (PLDs), digital signal processors (DSPs), field programmable gate arrays (FPGAs), logic gates, registers, semiconductor devices, chips, microchips, chipsets, etc. The local interface may include, for example, one or more buses or other wired or wireless connections, controllers, buffers (caches), drivers, repeaters, receivers, etc., to enable appropriate communication between the hardware components. A processor is a hardware device for executing software, particularly software stored in a memory. A processor can be any custom or commercially available processor, a central processing unit (CPU), an auxiliary processor among several processors associated with a computer, a semiconductor-based microprocessor (e.g., in the form of a microchip or chipset), a microprocessor, or generally any device for executing software instructions. A processor can also represent a distributed processing architecture. I / O devices can include input devices such as keyboards, mice, scanners, microphones, touch screens, interfaces for various medical devices and / or laboratory equipment, bar code readers, styluses, laser readers, radio frequency device readers, and the like. Additionally, I / O devices can also include output devices such as printers, bar code printers, displays, and the like.Finally, I / O devices may further include devices that communicate as both input and output, such as modulators / demodulators (modems; for accessing another device, system, or network), radio frequency (RF) or other transceivers, telephone interfaces, bridges, routers, etc.

[0076] Examples of software may include software components, programs, applications, computer programs, application programs, system programs, machine programs, operating system software, middleware, firmware, software modules, routines, subroutines, functions, methods, procedures, software interfaces, application program interfaces (APIs), instruction sets, opcodes, computer code, code segments, computer code segments, words, values, symbols, or any combination thereof. The software in the memory may include one or more separate programs that may include an ordered list of executable instructions for implementing logical functions. The software in the memory may include any suitable custom or commercially available operating system (O / S) that may control the execution of other computer programs, such as the system for identifying data streams according to the present teachings, as well as systems, and provides scheduling, input / output control, file and data management, memory management, communication control, and the like.

[0077] According to various exemplary embodiments, one or more features of any one or more of the teachings and / or exemplary embodiments discussed above may be performed or implemented using a suitably configured and / or programmed non-transitory machine-readable medium or article that may store instructions or sets of instructions that, when executed by a machine, may cause the machine to perform the methods and / or operations according to the exemplary embodiments. Such a machine may include, for example, any suitable processing platform, computing platform, computing device, processing device, computing system, processing system, computer, processor, scientific or laboratory equipment, etc., and may be implemented using any suitable combination of hardware and / or software. A machine-readable medium or article may include, for example, any suitable type of memory unit, memory device, memory article, memory medium, storage device, storage article, storage medium, and / or storage unit, such as memory, removable or non-removable media, erasable or non-erasable media, writable or rewritable media, digital or analog media, hard disk, floppy disk, read-only memory compact disc (CD-ROM), recordable compact disc (CD-R), rewriteable compact disc (CD-RW), optical disk, magnetic media, magneto-optical media, removable memory cards or disks, various types of Digital Versatile Disc (DVD), tape, cassette, etc., including any medium suitable for use in a computer. Memory may include any one or combination of volatile memory elements (e.g., random access memory (RAM, e.g., DRAM, SRAM, SDRAM, etc.)) and non-volatile memory elements (e.g., ROM, EPROM, EEROM, flash memory, hard drives, tapes, CDROM, etc.). Additionally, the memory may incorporate electrical, magnetic, optical, and / or other types of storage media.The memory may have a distributed architecture where various components are located remotely from one another but are still accessed by the processor. The instructions may include any suitable type of code, such as source code, compiled code, interpreted code, executable code, static code, dynamic code, encrypted code, etc., implemented using any suitable high-level, low-level, object-oriented, visual, compiled, and / or interpreted programming language.

[0078] According to various exemplary embodiments, one or more features of any one or more of the teachings and / or exemplary embodiments discussed above may be performed or implemented at least in part using distributed, cluster, remote, or cloud computing resources.

[0079] According to various exemplary embodiments, one or more features of any one or more of the teachings and / or exemplary embodiments discussed above may be performed or implemented using a source program, an executable program (object code), a script, or any other entity including a set of instructions to be performed. If a source program, the program may be translated via a compiler, assembler, interpreter, etc., which may or may not be included in memory, to communicate with an O / S and operate normally. The instructions may be written using (a) an object-oriented programming language with classes of data and methods, or (b) a procedural programming language with routines, subroutines, and / or functions, which may include, for example, C, C++, R, Pascal, Basic, Fortran, Cobol, Perl, Java, and Ada.

[0080] According to various exemplary embodiments, one or more of the exemplary embodiments discussed above may include transmitting, displaying, storing, printing, or outputting to a user interface device, a computer readable storage medium, a local computer system, or a remote computer system, any information, signals, data, and / or intermediate or final results generated, accessed, or used by such exemplary embodiments. Such transmitted, displayed, stored, printed, or outputted information may take the form of, for example, searchable and / or filterable listings of runs and reports, images, tables, charts, graphs, spreadsheets, correlations, arrays, and combinations thereof. EXAMPLES

[0081] Example 1 is a method for analyzing cells disposed on a sensor array surface of a ChemFET sensor array device, the method including: flowing a solution having a step change in pH across the sensor array surface, where a plurality of ChemFET sensors of the ChemFET sensor array generate a plurality of signals in response to the step change in pH of the flowed solution to generate electroscopic image data; acquiring a plurality of frames of electroscopic image data during an acquisition time interval, where each frame corresponds to a signal sample of a plurality of signals generated by the ChemFET sensor array measured at a sampling time during the acquisition time interval, where each frame includes a plurality of pixels, where a given pixel in the frame corresponds to a signal sample from a given sensor in the ChemFET sensor array; and segmenting the electroscopic image data into one or more cell regions corresponding to the location of cells on the sensor array surface and one or more background regions corresponding to areas on the sensor array that do not have cells based on characteristics of the signal samples generated in response to the step change in pH of the flowed solution.

[0082] Example 2 includes the subject matter of example 1, further specifying that each frame comprises a number of tiles, each tile comprising a subset of pixels of the frame.

[0083] Example 3 includes the subject matter of example 1, further specifying that the segmenting step further includes subtracting a minimum pixel value from a maximum pixel value for each pixel location during the acquisition time interval to form a peak-to-peak image.

[0084] Example 4 includes the subject matter of example 3, further specifying that the segmenting step further includes calculating an average value of pixels in the peak-to-peak image.

[0085] Example 5 includes the subject matter of example 4, further specifying that the segmenting step further includes subtracting an average value from each pixel value in the peak-to-peak image to form a difference image.

[0086] Example 6 includes the subject matter of example 5, further specifying that the segmenting step further includes calculating an absolute value for each pixel in the difference image.

[0087] Example 7 includes the subject matter of example 2, further specifying that the segmenting step further includes calculating average values ​​of pixels in the tile at each sampling time within the acquisition time interval to generate an array of average values ​​corresponding to each tile.

[0088] Example 8 includes the subject matter of example 2, further specifying that the segmenting step further includes dividing the tile into a plurality of subtiles.

[0089] Example 9 includes the subject matter of example 8, and further specifies that the segmenting step further includes calculating average values ​​of pixels within the subtile at each sampling time within the acquisition time interval to generate an array of average values ​​corresponding to each subtile.

[0090] Example 10 includes the subject matter of example 9, further specifying that the segmenting step further includes determining a frame index corresponding to a maximum average value in the array of average values.

[0091] Example 11 includes the subject matter of example 2, further specifying that the segmenting step further includes calculating an average over an acquisition time interval of each pixel value in the tile to form a first average image.

[0092] Example 12 includes the subject matter of Example 11, and further specifies that the segmenting step further includes calculating an average of each pixel value in the tile over a number of initial frames during the acquisition time interval to form a second average image.

[0093] Example 13 includes the subject matter of example 12, further specifying that the segmenting step further includes subtracting the second average image from the first average image to form a time-averaged image.

[0094] Example 14 includes the subject matter of example 1, further specifying that the segmenting step further includes determining for each pixel location a maximum pixel value during the acquisition time interval.

[0095] Example 15 includes the subject matter of example 14, further specifying that the segmenting step further includes determining a frame index corresponding to a maximum pixel value.

[0096] Example 16 includes the subject matter of example 14, further specifying that the segmenting step further includes determining a frame index corresponding to 80% of the maximum pixel value.

[0097] Example 17 includes the subject matter of example 2, further specifying that the segmenting step further includes determining feature values ​​based on characteristics of signal samples corresponding to each pixel of the tile to form a CellFind image.

[0098] Example 18 includes the subject matter of example 17, further specifying that the segmenting step further includes applying a blur function to the CellFind image to form a blurred CellFind image.

[0099] Example 19 includes the subject matter of example 18, further specifying that the blurring function includes applying a low pass filter to the CellFind image in the frequency domain.

[0100] Example 20 includes the subject matter of example 18, and further specifies that the segmenting step further includes applying a threshold to pixels of the blurred CellFind image to generate a coarse cell mask, where pixel values ​​greater than or equal to the threshold are assigned a value of 1 and pixel values ​​less than the threshold are assigned a value of 0.

[0101] Example 21 includes the subject matter of Example 20, and further specifies that the segmenting step further includes filtering the coarse cell mask to select features of the coarse cell mask having a size of interest to form a filtered cell mask.

[0102] Example 22 includes the subject matter of Example 21, further providing that the size of the object of interest corresponds to the size of a cell.

[0103] Example 23 includes the subject matter of example 21, and further specifies that the segmenting step further includes applying a watershed algorithm to selected features of the filtered cell mask to classify pixels of the filtered cell mask into one or more initial cell regions and one or more initial background regions to form the cell mask.

[0104] Example 24 includes the subject matter of example 23, further specifying that the segmenting step further includes validating the cell mask based on a Pearson difference calculated between the cell average time series and the cell and background time series for nearest subregions within the cell mask.

[0105] Example 25 includes the subject matter of example 24, further specifying that the segmenting step further includes clustering the Pearson differences to determine cell regions and background regions for a final cell mask.

[0106] Example 26 includes the subject matter of example 2, further specifying that the segmenting step further includes determining a plurality of feature values ​​based on a plurality of different characteristics of the signal samples corresponding to each pixel of the tile to form a CellFind image.

[0107] Example 27 includes the subject matter of example 26, further specifying that the segmenting step further includes applying a blur function to each of the plurality of CellFind images to form a plurality of blurred CellFind images.

[0108] Example 28 includes the subject matter of example 27, and further specifies that the segmenting step further includes applying a respective threshold to each of the plurality of blurred CellFind images to generate a plurality of coarse cell masks, wherein pixel values ​​equal to or greater than the respective threshold are assigned a value of 1 and pixel values ​​less than the respective threshold are assigned a value of 0.

[0109] Example 29 includes the subject matter of example 28, further specifying that the segmenting step further includes calculating a weighted sum of the multiple coarse cell masks to form a weighted sum image.

[0110] Example 30 includes the subject matter of example 29, and further specifies that the segmenting step further includes applying a second threshold to the weighted sum image to form a consensus cell mask, wherein pixel values ​​greater than or equal to the second threshold are assigned a value of 1, and pixel values ​​less than the second threshold are assigned a value of 0.

[0111] Example 31 includes the subject matter of Example 30, and further specifies that the segmenting step further includes filtering the consensus cell mask to select features of the consensus cell mask having a size of interest to form a filtered cell mask.

[0112] Example 32 includes the subject matter of Example 31, further providing that the size of the object of interest corresponds to the size of a cell.

[0113] Example 33 includes the subject matter of example 31, and further specifies that the segmenting step further includes applying a watershed algorithm to selected features of the filtered cell mask to classify pixels of the filtered cell mask into one or more initial cell regions and one or more initial background regions to form the cell mask.

[0114] Example 34 includes the subject matter of example 33, further specifying that the segmenting step further includes validating the cell mask based on a Pearson difference calculated between the cell average time series and the cell and background time series for the nearest subregion within the cell mask.

[0115] Example 35 includes the subject matter of example 34, further specifying that the segmenting step further includes clustering the Pearson differences to determine cell regions and background regions for a final cell mask.

[0116] Example 36 is a system for analyzing cells disposed on a sensor array surface of a ChemFET sensor array device, comprising a processor and a data store communicatively coupled to the processor, the processor configured to execute instructions that, when executed by the processor, cause the system to acquire a plurality of frames of electroscopic image data during an acquisition time interval, the electroscopic image data being generated in response to flowing a solution having a step change in pH across the sensor array surface, a plurality of ChemFET sensors of the ChemFET sensor array generating a plurality of signals in response to the step change in pH of the flowed solution to generate the electroscopic image data. and acquiring electroscopic image data based on characteristics of the signal samples generated in response to a step change in pH of the flowed solution into one or more cell regions corresponding to locations of cells on the sensor array surface and one or more background regions corresponding to areas on the sensor array that do not have cells, the electroscopic image data including a plurality of frames, each frame corresponding to signal samples of a plurality of signals generated by the ChemFET sensor array measured at sampling times during an acquisition time interval, where each frame includes a plurality of pixels, a given pixel in the frame corresponding to a signal sample from a given sensor in the ChemFET sensor array.

[0117] Example 37 includes the subject matter of example 36, further specifying that each frame comprises a plurality of tiles, each tile comprising a subset of pixels of the frame.

[0118] Example 38 includes the subject matter of example 36, further specifying that the segmenting step further includes subtracting a minimum pixel value from a maximum pixel value for each pixel location during the acquisition time interval to form a peak-to-peak image.

[0119] Example 39 includes the subject matter of example 38, further specifying that the segmenting step further includes calculating an average value of pixels in the peak-to-peak image.

[0120] Example 40 includes the subject matter of example 39, further specifying that the segmenting step further includes subtracting an average value from each pixel value in the peak-to-peak image to form a difference image.

[0121] Example 41 includes the subject matter of example 40, further specifying that the segmenting step further includes calculating an absolute value for each pixel in the difference image.

[0122] Example 42 includes the subject matter of example 37, further specifying that the segmenting step further includes calculating average values ​​of pixels in the tile at each sampling time within the acquisition time interval to generate an array of average values ​​corresponding to each tile.

[0123] Example 43 includes the subject matter of example 37, further specifying that the segmenting step further includes dividing the tile into a plurality of subtiles.

[0124] Example 44 includes the subject matter of Example 43, and further specifies that the segmenting step further includes calculating average values ​​of pixels within the subtile at each sampling time within the acquisition time interval to generate an array of average values ​​corresponding to each subtile.

[0125] Example 45 includes the subject matter of example 44, further specifying that the segmenting step further includes determining a frame index corresponding to a maximum average value in the array of average values.

[0126] Example 46 includes the subject matter of example 37, further specifying that the segmenting step further includes calculating an average of each pixel value in the tile over an acquisition time interval to form a first average image.

[0127] Example 47 includes the subject matter of Example 46, and further specifies that the segmenting step further includes calculating an average of each pixel value in the tile over multiple initial frames during the acquisition time interval to form a second average image.

[0128] Example 48 includes the subject matter of example 47, further specifying that the segmenting step further includes subtracting the second average image from the first average image to form a time-averaged image.

[0129] Example 49 includes the subject matter of example 36, further specifying that the segmenting step further includes determining a maximum pixel value during the acquisition time interval for each pixel location.

[0130] Example 50 includes the subject matter of example 49, further specifying that the segmenting step further includes determining a frame index corresponding to a maximum pixel value.

[0131] Example 51 includes the subject matter of example 49, further specifying that the segmenting step further includes determining a frame index corresponding to 80% of the maximum pixel value.

[0132] Example 52 includes the subject matter of Example 37, and further specifies that the segmenting step further includes determining feature values ​​based on characteristics of signal samples corresponding to each pixel of the tile to form a CellFind image.

[0133] Example 53 includes the subject matter of example 52, further specifying that the segmenting step further includes applying a blur function to the CellFind image to form a blurred CellFind image.

[0134] Example 54 includes the subject matter of example 53, further specifying that the blurring function includes applying a low pass filter to the CellFind image in the frequency domain.

[0135] Example 55 includes the subject matter of Example 53, and further specifies that the segmenting step further includes applying a threshold to pixels of the blurred CellFind image to generate a coarse cell mask, where pixel values ​​greater than or equal to the threshold are assigned a value of 1 and pixel values ​​less than the threshold are assigned a value of 0.

[0136] Example 56 includes the subject matter of Example 55, and further specifies that the segmenting step further includes filtering the coarse cell mask to select features of the coarse cell mask having a size of interest to form a filtered cell mask.

[0137] Example 57 includes the subject matter of Example 56, further providing that the size of the object of interest corresponds to the size of a cell.

[0138] Example 58 includes the subject matter of example 56, and further specifies that the segmenting step further includes applying a watershed algorithm to selected features of the filtered cell mask to classify pixels of the filtered cell mask into one or more initial cell regions and one or more initial background regions to form the cell mask.

[0139] Example 59 includes the subject matter of example 58, further specifying that the segmenting step further includes validating the cell mask based on a Pearson difference calculated between the cell average time series and the cell and background time series for the nearest subregion within the cell mask.

[0140] Example 60 includes the subject matter of example 59, further specifying that the segmenting step further includes clustering the Pearson differences to determine cell regions and background regions for a final cell mask.

[0141] Example 61 includes the subject matter of Example 37, and further specifies that the segmenting step further includes determining a plurality of feature values ​​based on a plurality of different characteristics of the signal samples corresponding to each pixel of the tile to form a CellFind image.

[0142] Example 62 includes the subject matter of example 61, further specifying that the segmenting step further includes applying a blur function to each of the plurality of CellFind images to form a plurality of blurred CellFind images.

[0143] Example 63 includes the subject matter of Example 62, and further specifies that the segmenting step further includes applying a respective threshold to each of the plurality of blurred CellFind images to generate a plurality of coarse cell masks, wherein pixel values ​​greater than or equal to the respective threshold are assigned a value of 1, and pixel values ​​less than the respective threshold are assigned a value of 0.

[0144] Example 64 includes the subject matter of example 63, further specifying that the segmenting step further includes calculating a weighted sum of the multiple coarse cell masks to form a weighted sum image.

[0145] Example 65 includes the subject matter of Example 64, and further specifies that the segmenting step further includes applying a second threshold to the weighted sum image to form a consensus cell mask, wherein pixel values ​​greater than or equal to the second threshold are assigned a value of 1, and pixel values ​​less than the second threshold are assigned a value of 0.

[0146] Example 66 includes the subject matter of Example 65, and further specifies that the segmenting step further includes filtering the consensus cell mask to select features of the consensus cell mask having a size of interest to form a filtered cell mask.

[0147] Example 67 includes the subject matter of Example 66, further providing that the size of the object of interest corresponds to the size of a cell.

[0148] Example 68 includes the subject matter of Example 66, and further specifies that the segmenting step further includes applying a watershed algorithm to selected features of the filtered cell mask to classify pixels of the filtered cell mask into one or more initial cell regions and one or more initial background regions to form the cell mask.

[0149] Example 69 includes the subject matter of example 68, further specifying that the segmenting step further includes validating the cell mask based on a Pearson difference calculated between the cell average time series and the cell and background time series for the nearest subregion within the cell mask.

[0150] Example 70 includes the subject matter of example 69, further specifying that the segmenting step further includes clustering the Pearson differences to determine cell regions and background regions for a final cell mask.

[0151] Example 71 is a non-transitory computer readable medium storing instructions that, when executed by a computer, cause the computer to perform a method for analyzing cells disposed on a sensor array surface of a ChemFET sensor array device, the method comprising acquiring a plurality of frames of electroscopic image data during an acquisition time interval, the electroscopic image data being generated in response to flowing a solution having a step change in pH across the sensor array surface, a plurality of ChemFET sensors of the ChemFET sensor array generating a plurality of signals in response to the step change in pH of the flowed solution to generate the electroscopic image data, each frame comprising: and acquiring electroscopic image data based on characteristics of the signal samples generated in response to a step change in pH of the flowed solution into one or more cell regions corresponding to locations of cells on the sensor array surface and one or more background regions corresponding to areas on the sensor array that do not have cells, wherein each frame includes a plurality of pixels, with a given pixel in the frame corresponding to a signal sample from a given sensor in the ChemFET sensor array measured at sampling times.

[0152] Example 72 includes the subject matter of example 71, further specifying that each frame comprises a plurality of tiles, each tile comprising a subset of pixels of the frame.

[0153] Example 73 includes the subject matter of Example 71, and further specifies that the segmenting step further includes subtracting a minimum pixel value from a maximum pixel value for each pixel location during the acquisition time interval to form a peak-to-peak image.

[0154] Example 74 includes the subject matter of example 73, further specifying that the segmenting step further includes calculating an average value of pixels in the peak-to-peak image.

[0155] Example 75 includes the subject matter of example 74, further specifying that the segmenting step further includes subtracting an average value from each pixel value in the peak-to-peak image to form a difference image.

[0156] Example 76 includes the subject matter of example 75, further specifying that the segmenting step further includes calculating an absolute value for each pixel in the difference image.

[0157] Example 77 includes the subject matter of Example 72, and further specifies that the segmenting step further includes calculating average values ​​of pixels in the tile at each sampling time within the acquisition time interval to generate an array of average values ​​corresponding to each tile.

[0158] Example 78 includes the subject matter of Example 72, further specifying that the segmenting step further includes dividing the tile into a plurality of subtiles.

[0159] Example 79 includes the subject matter of Example 78, and further specifies that the segmenting step further includes calculating average values ​​of pixels within the subtile at each sampling time within the acquisition time interval to generate an array of average values ​​corresponding to each subtile.

[0160] Example 80 includes the subject matter of example 79, further specifying that the segmenting step further includes determining a frame index corresponding to a maximum average value in the array of average values.

[0161] Example 81 includes the subject matter of Example 72, and further specifies that the segmenting step further includes calculating an average of each pixel value in the tile over an acquisition time interval to form a first average image.

[0162] Example 82 includes the subject matter of Example 81, and further specifies that the segmenting step further includes calculating an average of each pixel value in the tile over multiple initial frames during the acquisition time interval to form a second average image.

[0163] Example 83 includes the subject matter of example 82, further specifying that the segmenting step further includes subtracting the second average image from the first average image to form a time-averaged image.

[0164] Example 84 includes the subject matter of example 71, further specifying that the segmenting step further includes determining a maximum pixel value during the acquisition time interval for each pixel location.

[0165] Example 85 includes the subject matter of Example 84, further specifying that the segmenting step further includes determining a frame index corresponding to a maximum pixel value.

[0166] Example 86 includes the subject matter of example 84, further specifying that the segmenting step further includes determining a frame index corresponding to 80% of the maximum pixel value.

[0167] Example 87 includes the subject matter of Example 72, and further specifies that the segmenting step further includes determining feature values ​​based on characteristics of signal samples corresponding to each pixel of the tile to form a CellFind image.

[0168] Example 88 includes the subject matter of example 87, further specifying that the segmenting step further includes applying a blur function to the CellFind image to form a blurred CellFind image.

[0169] Example 89 includes the subject matter of example 88, further specifying that the blurring function includes applying a low pass filter to the CellFind image in the frequency domain.

[0170] Example 90 includes the subject matter of Example 88, and further specifies that the segmenting step further includes applying a threshold to pixels of the blurred CellFind image to generate a coarse cell mask, wherein pixel values ​​greater than or equal to the threshold are assigned a value of 1 and pixel values ​​less than the threshold are assigned a value of 0.

[0171] Example 91 includes the subject matter of Example 90, and further specifies that the segmenting step further includes filtering the coarse cell mask to select features of the coarse cell mask having a size of interest to form a filtered cell mask.

[0172] Example 92 includes the subject matter of Example 91, further providing that the size of the object of interest corresponds to the size of a cell.

[0173] Example 93 includes the subject matter of Example 91, and further specifies that the segmenting step further includes applying a watershed algorithm to selected features of the filtered cell mask to classify pixels of the filtered cell mask into one or more initial cell regions and one or more initial background regions to form the cell mask.

[0174] Example 94 includes the subject matter of Example 93, and further specifies that the segmenting step further includes validating the cell mask based on a Pearson difference calculated between the cell average time series and the cell and background time series for the nearest subregion within the cell mask.

[0175] Example 95 includes the subject matter of example 94, further specifying that the segmenting step further includes clustering the Pearson differences to determine cell regions and background regions for a final cell mask.

[0176] Example 96 includes the subject matter of Example 72, and further specifies that the segmenting step further includes determining a plurality of feature values ​​based on a plurality of different characteristics of the signal samples corresponding to each pixel of the tile to form a CellFind image.

[0177] Example 97 includes the subject matter of example 96, further specifying that the segmenting step further includes applying a blur function to each of the plurality of CellFind images to form a plurality of blurred CellFind images.

[0178] Example 98 includes the subject matter of Example 97, and further specifies that the segmenting step further includes applying a respective threshold to each of the plurality of blurred CellFind images to generate a plurality of coarse cell masks, wherein pixel values ​​greater than or equal to the respective threshold are assigned a value of 1, and pixel values ​​less than the respective threshold are assigned a value of 0.

[0179] Example 99 includes the subject matter of example 98, further specifying that the segmenting step further includes calculating a weighted sum of the multiple coarse cell masks to form a weighted sum image.

[0180] Example 100 includes the subject matter of Example 99, and further specifies that the segmenting step further includes applying a second threshold to the weighted sum image to form a consensus cell mask, and pixel values ​​greater than or equal to the second threshold are assigned a value of 1, and pixel values ​​less than the second threshold are assigned a value of 0.

[0181] Example 101 includes the subject matter of Example 100, and further specifies that the segmenting step further includes filtering the consensus cell mask to select features of the consensus cell mask having a size of interest to form a filtered cell mask.

[0182] Example 102 includes the subject matter of Example 101, further providing that the size of the object of interest corresponds to the size of a cell.

[0183] Example 103 includes the subject matter of Example 101, and further specifies that the segmenting step further includes applying a watershed algorithm to selected features of the filtered cell mask to classify pixels of the filtered cell mask into one or more initial cell regions and one or more initial background regions to form the cell mask.

[0184] Example 104 includes the subject matter of example 103, and further specifies that the segmenting step further includes validating the cell mask based on a Pearson difference calculated between the cell average time series and the cell and background time series for the nearest subregion within the cell mask.

[0185] Example 105 includes the subject matter of example 104, further specifying that the segmenting step further includes clustering the Pearson differences to determine cell regions and background regions for a final cell mask.

[0186] Example A1 is a method for analyzing cells disposed on a sensor array surface of a ChemFET sensor array device, the method including: flowing a solution having a step change in pH across the sensor array surface, where a plurality of ChemFET sensors of the ChemFET sensor array generate a plurality of signals in response to the step change in pH of the flowed solution to generate electroscopic image data; acquiring a plurality of frames of electroscopic image data during an acquisition time interval, where each frame corresponds to a signal sample of a plurality of signals generated by the ChemFET sensor array measured at a sampling time during the acquisition time interval, where each frame includes a plurality of pixels, where a given pixel in the frame corresponds to a signal sample from a given sensor in the ChemFET sensor array; and segmenting the electroscopic image data into one or more cell regions corresponding to the location of cells on the sensor array surface and one or more background regions corresponding to areas on the sensor array that do not have cells based on characteristics of the signal samples generated in response to the step change in pH of the flowed solution.

[0187] Example A2 includes the subject matter of example A1, further specifying that each frame comprises a plurality of tiles, each tile comprising a subset of pixels of the frame.

[0188] Example A3 includes the subject matter of example A1, and further specifies that the segmenting step further includes subtracting a minimum pixel value from a maximum pixel value for each pixel location during the acquisition time interval to form a peak-to-peak image.

[0189] Example A4 includes the subject matter of example A3, and further specifies that the segmenting step further includes calculating an average value of pixels in the peak-to-peak image, subtracting the average value from each pixel value in the peak-to-peak image to form a difference image, and calculating an absolute value of each pixel in the difference image.

[0190] Example A5 includes the subject matter of example A2, and further specifies that the segmenting step further includes determining feature values ​​based on characteristics of signal samples corresponding to each pixel of the tile to form a CellFind image.

[0191] Example A6 includes the subject matter of example A5, and further specifies that the segmenting step further includes applying a blur function to the CellFind image to form a blurred CellFind image, the blur function further including applying a low-pass filter to the CellFind image.

[0192] Example A7 includes the subject matter of Example A6, and further specifies that the segmenting step further includes applying a threshold to pixels of the blurred CellFind image to generate a coarse cell mask, where pixel values ​​greater than or equal to the threshold are assigned a value of 1 and pixel values ​​less than the threshold are assigned a value of 0.

[0193] Example A8 includes the subject matter of Example A7, and further specifies that the segmenting step further includes filtering the coarse cell mask to select features of the coarse cell mask having a size of interest to form a filtered cell mask.

[0194] Example A9 includes the subject matter of Example A8, and further specifies that the segmenting step further includes applying a watershed algorithm to selected features of the filtered cell mask to classify pixels of the filtered cell mask into one or more initial cell regions and one or more initial background regions to form the cell mask.

[0195] Example A10 includes the subject matter of Example A9, and further specifies that the segmenting step further includes validating the cell mask based on Pearson differences calculated between the cell average time series and the cell and background time series for nearest subregions within the cell mask, and clustering the Pearson differences to determine cell regions and background regions for the final cell mask.

[0196] Example A11 is a system for analyzing cells disposed on a sensor array surface of a ChemFET sensor array device, comprising: a processor; and a data store communicatively coupled to the processor, the processor configured to execute instructions that, when executed by the processor, cause the system to acquire a plurality of frames of electroscopic image data during an acquisition time interval, the electroscopic image data being generated in response to flowing a solution having a step change in pH across the sensor array surface, a plurality of ChemFET sensors of the ChemFET sensor array generating a plurality of signals in response to the step change in pH of the flowed solution to store the electroscopic image data. and acquiring electroscopic image data, each frame corresponding to signal samples of a plurality of signals generated by the ChemFET sensor array measured at sampling times during an acquisition time interval, each frame including a plurality of pixels, a given pixel in a frame corresponding to a signal sample from a given sensor in the ChemFET sensor array; and segmenting the electroscopic image data into one or more cell regions corresponding to locations of cells on the sensor array surface and one or more background regions corresponding to areas on the sensor array that do not have cells based on characteristics of the signal samples generated in response to a step change in pH of the flowed solution.

[0197] Example A12 includes the subject matter of example A11, further specifying that each frame comprises a plurality of tiles, each tile comprising a subset of pixels of the frame.

[0198] Example A13 includes the subject matter of example A11, and further specifies that the segmenting step further includes subtracting a minimum pixel value from a maximum pixel value for each pixel location during the acquisition time interval to form a peak-to-peak image.

[0199] Example A14 includes the subject matter of example A13, and further specifies that the segmenting step further includes calculating an average value of pixels in the peak-to-peak image, subtracting the average value from each pixel value in the peak-to-peak image to form a difference image, and calculating an absolute value of each pixel in the difference image.

[0200] Example A15 includes the subject matter of example A12, and further specifies that the segmenting step further includes determining feature values ​​based on characteristics of signal samples corresponding to each pixel of the tile to form a CellFind image.

[0201] Example A16 includes the subject matter of example A15, and further specifies that the segmenting step further includes applying a blur function to the CellFind image to form a blurred CellFind image, the blur function further including applying a low-pass filter to the CellFind image.

[0202] Example A17 includes the subject matter of example A16, and further specifies that the segmenting step further includes applying a threshold to pixels of the blurred CellFind image to generate a coarse cell mask, wherein pixel values ​​greater than or equal to the threshold are assigned a value of 1 and pixel values ​​less than the threshold are assigned a value of 0.

[0203] Example A18 includes the subject matter of Example A17, and further specifies that the segmenting step further includes filtering the coarse cell mask to select features of the coarse cell mask having a size of interest to form a filtered cell mask.

[0204] Example A19 includes the subject matter of Example A18, and further specifies that the segmenting step further includes applying a watershed algorithm to selected features of the filtered cell mask to classify pixels of the filtered cell mask into one or more initial cell regions and one or more initial background regions to form the cell mask.

[0205] Example A20 includes the subject matter of example A19, and further specifies that the segmenting step further includes validating the cell mask based on Pearson differences calculated between the cell average time series and the cell and background time series for nearest subregions within the cell mask, and clustering the Pearson differences to determine cell regions and background regions for the final cell mask.

[0206] Example A21 is a non-transitory computer readable medium storing instructions that, when executed by a computer, cause the computer to perform a method for analyzing cells disposed on a sensor array surface of a ChemFET sensor array device, the method comprising acquiring a plurality of frames of electroscopic image data during an acquisition time interval, the electroscopic image data being generated in response to flowing a solution having a step change in pH across the sensor array surface, a plurality of ChemFET sensors of the ChemFET sensor array generating a plurality of signals in response to the step change in pH of the flowed solution to generate the electroscopic image data, each frame being within an acquisition time interval. a frame corresponding to signal samples of a plurality of signals generated by a ChemFET sensor array measured at sampling times during an interval, each frame including a plurality of pixels, a given pixel in the frame corresponding to a signal sample from a given sensor in the ChemFET sensor array; and segmenting the electroscopic image data into one or more cell regions corresponding to locations of cells on the sensor array surface and one or more background regions corresponding to areas on the sensor array that do not have cells based on characteristics of the signal samples generated in response to a step change in pH of the flowed solution.

[0207] Example A22 includes the subject matter of example A21, further specifying that each frame comprises a plurality of tiles, each tile comprising a subset of pixels of the frame.

[0208] Example A23 includes the subject matter of example A21, and further specifies that the segmenting step further includes subtracting a minimum pixel value from a maximum pixel value for each pixel location during the acquisition time interval to form a peak-to-peak image.

[0209] Example A24 includes the subject matter of example A23, and further specifies that the segmenting step further includes calculating an average value of pixels in the peak-to-peak image, subtracting the average value from each pixel value in the peak-to-peak image to form a difference image, and calculating an absolute value of each pixel in the difference image.

[0210] Example A25 includes the subject matter of example A22, and further specifies that the segmenting step further includes determining feature values ​​based on characteristics of signal samples corresponding to each pixel of the tile to form a CellFind image.

[0211] Example A26 includes the subject matter of example A25, and further specifies that the segmenting step further includes applying a blurring function to the CellFind image to form a blurred CellFind image, the blurring function further including applying a low-pass filter to the CellFind image.

[0212] Example A27 includes the subject matter of example A26, and further specifies that the segmenting step further includes applying a threshold to pixels of the blurred CellFind image to generate a coarse cell mask, wherein pixel values ​​greater than or equal to the threshold are assigned a value of 1 and pixel values ​​less than the threshold are assigned a value of 0.

[0213] Example A28 includes the subject matter of example A27, and further specifies that the segmenting step further includes filtering the coarse cell mask to select features of the coarse cell mask having a size of interest to form a filtered cell mask.

[0214] Example A29 includes the subject matter of example A28, and further specifies that the segmenting step further includes applying a watershed algorithm to selected features of the filtered cell mask to classify pixels of the filtered cell mask into one or more initial cell regions and one or more initial background regions to form the cell mask.

[0215] Example A30 includes the subject matter of example A29, and further specifies that the segmenting step further includes validating the cell mask based on Pearson differences calculated between the cell average time series and the cell and background time series for nearest subregions within the cell mask, and clustering the Pearson differences to determine cell regions and background regions for the final cell mask.

Claims

1. 1. A method for analyzing cells disposed on a sensor array surface of a ChemFET sensor array device, comprising: flowing a solution having a step change in pH across a surface of the sensor array, wherein a plurality of ChemFET sensors of the ChemFET sensor array generate a plurality of signals in response to the step change in pH of the flowed solution to generate electroscopic image data; acquiring a plurality of frames of the electroscopic image data during an acquisition time interval, each frame corresponding to a signal sample of the plurality of signals generated by the ChemFET sensor array measured at a sampling time during the acquisition time interval, each frame including a plurality of pixels, a given pixel in the frame corresponding to a signal sample from a given sensor in the ChemFET sensor array; and segmenting the electroscopic image data into one or more cell regions corresponding to the location of the cells on the sensor array surface and one or more background regions corresponding to areas on the sensor array that do not have cells based on characteristics of the signal samples generated in response to a step change in the pH of the flowed solution.

2. The method of claim 1 , wherein each frame comprises a number of tiles, each tile comprising a subset of pixels of the frame.

3. The method of claim 1 , wherein the segmenting step further comprises: subtracting a minimum pixel value from a maximum pixel value for each pixel location during the acquisition time interval to form a peak-to-peak image.

4. The segmenting step further comprises: calculating an average value of pixels in the peak-to-peak image; subtracting the average value from each pixel value in the peak-to-peak image to form a difference image; The method of claim 3 further comprising: calculating an absolute value of each pixel in the difference image.

5. The method of claim 2 , wherein the segmenting step further comprises determining feature values ​​based on characteristics of the signal samples corresponding to each pixel of the tile to form a CellFind image.

6. 6. The method of claim 5, wherein the segmenting step further comprises applying a blurring function to the CellFind image to form a blurred CellFind image, the blurring function comprising applying a low pass filter to the CellFind image.

7. 7. The method of claim 6, wherein the segmenting step further comprises applying a threshold to pixels of the blurred CellFind image to generate a coarse cell mask, where pixel values ​​greater than or equal to the threshold are assigned a value of 1 and pixel values ​​less than the threshold are assigned a value of 0.

8. 8. The method of claim 7, wherein the segmenting step further comprises filtering the coarse cell mask to select features of the coarse cell mask having a size of interest to form a filtered cell mask.

9. 9. The method of claim 8, wherein the segmenting step further comprises applying a watershed algorithm to the selected features of the filtered cell mask to classify pixels of the filtered cell mask into one or more initial cell regions and one or more initial background regions to form a cell mask.

10. The segmenting step further comprises: validating the cell mask based on a Pearson difference calculated between the cell mean time series and the cell and background time series for nearest sub-regions within the cell mask; 10. The method of claim 9, further comprising: clustering the Pearson differences to determine the cell regions and the background regions for a final cell mask.

11. 1. A system for analyzing cells disposed on a sensor array surface of a ChemFET sensor array device, comprising: a processor; and a data store communicatively coupled to the processor, the processor configured to execute instructions that, when executed by the processor, provide the system with: acquiring a plurality of frames of electroscopic image data during an acquisition time interval, the electroscopic image data being generated in response to flowing a solution having a step change in pH across a surface of the sensor array, a plurality of ChemFET sensors of the ChemFET sensor array generating a plurality of signals in response to the step change in pH of the flowed solution to generate the electroscopic image data, each frame corresponding to a signal sample of the plurality of signals generated by the ChemFET sensor array measured at a sampling time during the acquisition time interval, each frame including a plurality of pixels, a given pixel in the frame corresponding to a signal sample from a given sensor in the ChemFET sensor array; and segmenting the electroscopic image data into one or more cell regions corresponding to the location of the cells on the sensor array surface and one or more background regions corresponding to areas on the sensor array that do not have cells based on characteristics of the signal samples generated in response to a step change in pH of the flowed solution.

12. The system of claim 11 , wherein each frame comprises a number of tiles, each tile comprising a subset of pixels of the frame.

13. 12. The system of claim 11, wherein the segmenting step further comprises: subtracting a minimum pixel value from a maximum pixel value for each pixel location during the acquisition time interval to form a peak-to-peak image.

14. The segmenting step further comprises: calculating an average value of pixels in the peak-to-peak image; subtracting the average value from each pixel value in the peak-to-peak image to form a difference image; The system of claim 13 further comprising: calculating an absolute value of each pixel in the difference image.

15. The system of claim 12 , wherein the segmenting step further comprises determining feature values ​​based on the characteristics of the signal samples corresponding to each pixel of the tile to form a CellFind image.

16. 16. The system of claim 15, wherein the segmenting step further comprises applying a blurring function to the CellFind image to form a blurred CellFind image, the blurring function comprising applying a low pass filter to the CellFind image.

17. 17. The system of claim 16, wherein the segmenting step further comprises applying a threshold to pixels of the blurred CellFind image to generate a coarse cell mask, where pixel values ​​greater than or equal to the threshold are assigned a value of 1 and pixel values ​​less than the threshold are assigned a value of 0.

18. 20. The system of claim 17, wherein the segmenting step further comprises filtering the coarse cell mask to select features of the coarse cell mask having a size of interest to form a filtered cell mask.

19. 20. The system of claim 18, wherein the segmenting step further comprises applying a watershed algorithm to the selected features of the filtered cell mask to classify pixels of the filtered cell mask into one or more initial cell regions and one or more initial background regions to form a cell mask.

20. The segmenting step further comprises: validating the cell mask based on a Pearson difference calculated between the cell mean time series and the cell and background time series for nearest sub-regions within the cell mask; 20. The system of claim 19, further comprising: clustering the Pearson differences to determine the cell regions and the background regions for a final cell mask.

Citation Information

Patent Citations

  • Method and apparatus for analyte measurement using a large-scale FET array

    JP2010513869A

  • Apparatus and method for carrying out electrochemical reactions

    JP2012528329A

  • Measurement device and method for screening with cells immobilized on electrodes

    US20050037438A1

  • Methods for sequencing individual nucleic acids under tension

    US20100035252A1

  • Cell analysis using chemfet sensor array-based systems

    US20200088676A1