Methods and systems for detecting abnormalities in a foot

Abstract

A method executed by a computer processor for detecting abnormalities in a foot is disclosed. The method involves receiving thermal images of a foot and applying these images to a pre-trained classifier to classify regions of the foot into categories based on temperature, including normal, low, and high temperature categories. The method further processes the thermal images to generate segmented images by isolating portions of the images corresponding to regions of the foot classified as low or high temperature. Additionally, the method calculates the area of each region classified as low or high temperature, providing quantitative data for further analysis and clinical assessment.

Description

[0001] Methods And Systems For Detecting Abnormalities In A Foot

[0002] TECHNICAL FIELD

[0003] [1] The present invention relates to systems and methods for detecting abnormalities in a foot. In particular, embodiments of the invention relate to detecting abnormalities in a foot for the purpose of early detection and treatment of diabetic foot ulcers (DFUs).

[0004] BACKGROUND

[0005] [2] Any references to methods, apparatus or documents of the prior art are not to be taken as constituting any evidence or admission that they formed, or form part of the common general knowledge.

[0006] [3] Diabetic foot ulcers (DFUs) are a serious complication associated with diabetes mellitus and represent a significant burden on healthcare systems worldwide. DFUs develop due to a combination of peripheral neuropathy and poor circulation, resulting in non-healing wounds prone to infection. These ulcers are classified into six stages based on severity:

[0007] Stage 0: No open lesions I Healed lesions - A stage of risk assessment or recovery;

[0008] Stage 1: Superficial ulcer - Affects only the upper layers of the skin, without deeper penetration;

[0009] Stage 2: Deep ulcer - Extends to underlying tissues, such as tendons, bones, or joints;

[0010] Stage 3: Deep ulcers with infected tissues - Complications arise due to secondary infections;Stage 4: Partial gangrene - Necrosis occurs in limited areas of the foot; and

[0011] Stage 5: Complete gangrene - Extensive necrosis, often requiring major amputation.

[0012] [4] Effective treatment depends heavily on detecting DFUs at earlier stages, particularly Stage 0 (prior to visible ulceration), when interventions are most successful in preventing disease progression. However, current diagnostic approaches are largely inadequate for early detection, relying on subjective methods such as visual inspection or sensory testing. These techniques often fail to identify subtle changes in circulation or inflammation, which are precursors to DFU development.

[0013] [5] Current methods for detecting and monitoring DFUs are predominantly reliant on subjective assessments, such as visual inspection and sensory testing. These methods are often performed after symptoms become visibly apparent (e.g. Stage 1 or later), such as open sores or discoloration, which indicates that the condition is already advanced. This reliance on late-stage detection reduces the window for effective intervention, increasing the risk of complications and poor patient outcomes.

[0014] [6] Additionally, the lack of objective, quantifiable metrics for assessing the extent or progression of DFUs presents a challenge for healthcare providers. Monitoring the effectiveness of treatments and tracking healing over time is often imprecise and varies significantly between practitioners, leading to inconsistent care.

[0015] [7] Existing diagnostic tools and methods, such as imaging technologies, are often inadequate, ineffective, expensive, invasive, or impractical for widespread use in resource-constrained environments. In particular, these tools may also fail todetect early-stage abnormalities that precede visible ulcer formation, further limiting their utility in preventing complications.

[0016] [8] There is a need for a diagnostic solution that enables early, objective, and reliable detection of abnormalities associated with DFUs, providing actionable insights to clinicians while being accessible for use in diverse clinical and remote settings.

[0017] SUMMARY OF INVENTION

[0018] [9] In an aspect, the invention provides a method executed by a computer processor for detecting abnormalities in a foot, the method comprising:

[0019] receiving thermal images of a foot;

[0020] applying the thermal images to a pre-trained classifier to classify regions of the foot into categories based on temperature, wherein the categories are one of: normal, low or high temperature; and

[0021] processing the thermal images to generate segmented images by isolating portions of the thermal images corresponding to the regions of the foot categorised as low or high temperature; and

[0022] calculating an area of each region of the foot categorised as low or high temperature.

[0023]

[0010] In another aspect, the invention provides a system for processing a thermal image to detecting abnormalities in a foot, the system comprising:

[0024] a computing device configured to perform the steps of:

[0025] receiving thermal images of a foot;applying the thermal images to a pre-trained classifier to classify regions of the foot into categories based on temperature, wherein the categories are one of: normal, low or high temperature; and

[0026] processing the thermal images to generate segmented images by isolating portions of the thermal images corresponding to the regions of the foot categorised as low or high temperature; and

[0027] calculating an area of each region of the foot categorised as low or high temperature.

[0028]

[0011] Preferably, the abnormal category comprises two sub-categories: low temperature and high temperature. Preferably, the classifier further classifies regions classified as abnormal as one of: low or high temperature.

[0029]

[0012] Preferably, the method comprises receiving thermal images of two feet, including a left foot and a right foot of a patient. Preferably, the method comprises the step of capturing thermal images of a foot or feet.

[0030]

[0013] Preferably, the regions are defined by a collection of pixels in the thermal images.

[0031]

[0014] Preferably, the step of applying the thermal images to the pre-trained classifier comprises applying predetermined temperature thresholds (standardised temperature thresholds) to classify regions of the foot, wherein temperatures within the range of 28°C to 34°C are classified as normal, temperatures below 28°C are classified as low, and temperatures above 34°C are classified as high.

[0032]

[0015] Preferably, the step of applying the thermal images to the pre-trained classifier comprises calculating a mean temperature of the foot and comparing each pixel’s temperature to the mean to identify deviations exceeding a predeterminedrelative threshold (relative temperature thresholds). Alternatively, or additionally, the step of applying the thermal images to the pre-trained classifier comprises calculating a mean temperature of a portion of the foot surrounding a pixel and comparing the pixel’s temperature to the mean to identify deviations exceeding a predetermined threshold. Preferably, the predetermined relative threshold is approximately ±2°C.

[0033]

[0016] Preferably, the method comprises the step of comparing corresponding regions of the thermal image of contralateral feet to identify asymmetrical temperature differences exceeding a predefined contralateral threshold. Preferably, the predefined contralateral threshold is approximately ±2.2°C.

[0034]

[0017] Preferably, the method comprises the step of adjusting the classification of one or more regions of the thermal image. More preferably, the method comprises the step of adjusting the classification of one or more regions of the thermal images identified as exceeding the predetermined relative threshold and / or the predetermined contralateral threshold. Preferably, the adjustment of the classification is based on a direction and magnitude of the deviation.

[0035]

[0018] Preferably, the segmented images include only the regions of the foot categorised as low or high temperature regions.

[0036]

[0019] Preferably, the computing device is configured to perform the step of displaying the segmented images and the thermal images.

[0037]

[0020] Preferably, the method comprises the step of receiving patient health data and processing the patient health data using a machine learning algorithm in combination with classifications of the thermal images of the foot to provide a risk assessment for DFU development.

[0021] In another aspect, the invention provides a method executed by a computer processor for monitoring abnormalities in a foot over time, the method comprising: receiving thermal images of a foot from a first imaging procedure and a subsequent imaging procedure;

[0038] applying the thermal images from each imaging procedure to a pre-trained classifier to classify regions of the foot into categories based on temperature, wherein the categories are one of: low, normal, and high temperatures;

[0039] generating segmented images for each imaging procedure by isolating portions of the thermal images corresponding to regions of the foot classified as low or high temperature;

[0040] comparing the segmented images from the first imaging procedure and the subsequent imaging procedure to identify changes between the thermal images from the first imaging procedure and thermal images from the subsequent imaging procedure; and

[0041] generating a report indicating the changes identified between the imaging procedures to assist in treatment planning.

[0042]

[0022] Preferably, the changes in the regions include changes in size and / or temperature of the regions.

[0043]

[0023] Preferably, the comparing step including comparing the segmented images from the first imaging procedure and the subsequent imaging procedure to identify changes between the thermal images from the first imaging procedure and thermal images from the subsequent imaging procedure in the regions classified as low or high temperature.

[0024] Preferably, the method includes calculating quantitative changes in the images. Preferably, the method includes calculating quantitative changes in the regions classified as low or high temperature.

[0044]

[0025] In another aspect, the invention provides a method executed by a computer processor for detecting abnormalities in a foot, the method comprising:

[0045] receiving thermal images of a foot;

[0046] applying the thermal images to a pre-trained classifier to classify regions of the foot into categories based on temperature, wherein the categories are one of: normal and abnormal; and

[0047] processing the thermal images to generate segmented images by isolating portions of the thermal images corresponding to the regions of the foot categorised as abnormal regions; and

[0048] calculating an area of each region of the foot categorised as abnormal regions.

[0049] BRIEF DESCRIPTION OF THE DRAWINGS

[0050]

[0026] Preferred features, embodiments and variations of the invention may be discerned from the following Detailed Description which provides sufficient information for those skilled in the art to perform the invention. The Detailed Description is not to be regarded as limiting the scope of the preceding Summary of the Invention in any way. The Detailed Description will make reference to a number of drawings as follows:

[0051] Figure 1 illustrates a system for detecting abnormalities in a foot according to an embodiment of the present invention;Figure 2 illustrates a method for detecting abnormalities in a foot according to an embodiment of the present invention;

[0052] Figure 3 illustrates a method for detecting abnormalities in a foot over time according to an embodiment of the present invention; and

[0053] Figures 4-6 illustrate examples of the thermal images that are classified and segmented.

[0054] DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS

[0055]

[0027] Figures 1 to 3 illustrate methods and systems for detecting abnormalities in a foot. The method is executed by a processor of a computing device 110 that is appropriately configured to do so.

[0056]

[0028] In one step, the method, at box 310, includes receiving thermal images of a foot. The method, at box 305, may also include capturing the thermal images but this step is optional.

[0057]

[0029] Thermal images are captured using a thermal imaging device 120 configured to capture thermal images. The thermal imaging device 120 is preferably a high-resolution scanner, such as an infrared scanner, imager or camera. This is known as infrared thermography (IRT). To measure the skin temperature in a particular area, IRT uses a non-invasive method with a specialised camera that captures long-infrared radiation from the electromagnetic spectrum (9,000-14,000 nanometres or 9-14 m). This produces thermograms (images) showing patterns of heat and blood flow on or near the surface of the body.

[0058]

[0030] An example of an image acquisition procedure is provided below.

[0059]

[0031] The imaging acquisition process begins with the preparation of a controlled environment to ensure the accuracy and consistency of the thermal imagescaptured. A suitable room 130 is selected to facilitate the comfortable placement and operation of the equipment and to accommodate the patient 140 and technician. The environmental conditions within the room are carefully regulated, with the room temperature maintained at 20°C ± 1°C, relative humidity set to 50%, and reflected temperature controlled at 20°C. The emissivity of the thermal imaging device is, in one embodiment, configured to 0.98, which is suitable for human skin. To minimize interference from external infrared radiation, windows and light sources may be shielded or covered as necessary.

[0060]

[0032] The patient is prepared by removing shoes and socks, allowing the foot temperature to stabilize over a period of time. In one embodiment, the period of time is approximately ten minutes. During this time, the patient is seated comfortably in a chair or lies on a medical bed, with their feet positioned for optimal imaging relative to the thermal imaging device. This stabilization period ensures consistent blood circulation and vessel temperature, enhancing the reliability of the thermal readings.

[0061]

[0033] The thermal imaging device 120 is positioned at a fixed distance of approximately one metre from the patient’s feet. The thermal imaging device 120 is calibrated appropriately to capture high-resolution thermal images, and its alignment may be checked to ensure accurate coverage of the target areas.

[0062]

[0034] To capture comprehensive thermal data, multiple images may be taken from different angles, including the plantar (sole), dorsal (top), medial (inner side), and lateral (outer side) views of the foot. The patient is instructed to remain as still as possible during the imaging process to minimize motion artifacts. Privacy is maintained throughout the procedure using an isolation wall or privacy screen 150 placed between the face of the patient and the thermal imaging device to shield the patient from view.

[0035] After capturing the images, the thermal data may be reviewed to verify image quality and ensure there are no distortions or artifacts. If the images do not meet the required quality standards, the imaging process may be repeated. Once verified, the images can be securely stored for further analysis and comparison across subsequent imaging procedures.

[0063]

[0036] This procedure ensures that the captured thermal images are of the highest quality, providing accurate data for the detection and monitoring of diabetic foot ulcers and other foot-related abnormalities.

[0064]

[0037] In another step illustrated at box 212, the thermal images may undergo preprocessing. The pre-processing may include normalisation and / or reduction of missing data.

[0065]

[0038] In another step, at box 215, the thermal images, whether they have been pre-processed or not, are applied to a pre-trained classifier to identify and classify regions of the foot into categories based on temperature. The categories include normal and abnormal temperatures, where abnormal temperatures are classified as either low or high temperatures.

[0066]

[0039] The regions may be made up of or defined by a collection of pixels in the thermal images.

[0067]

[0040] Temperature values may be determined and assigned on a pixel-by-pixel basis.

[0068]

[0041] These images may be classified using the Vision Transformer Al model into one of the three temperature categories: normal, low, and high.

[0069]

[0042] The integration of Al models specifically tailored for temperature classification and segmentation (to be described below) ensures objective and early detection of abnormalities, unlike traditional visual inspection methods that are subjective andoften delayed. The system is therefore able to detect any early signs of poor circulation of the blood, which will lead to DFU later.

[0070]

[0043] The classifications of low, normal and high are based on relative temperature deviations identified from images of each patient’s foot, combined with standardised thresholds established from a validated dataset of diabetic and non-diabetic individuals. This ensures both personalised and consistent evaluations.

[0071]

[0044] With reference to standardised thresholds, these can be understood from the following.

[0072] Normal Temperature

[0073]

[0045] No significant deviations were observed. The temperature distribution across the foot is uniform and within the expected physiological ranges. Foot temperatures typically range between 28°C and 34°C under normal physiological conditions. Therefore, temperatures within this range are classified as normal unless significant relative deviations are present.

[0074] Low Temperature

[0075]

[0046] Areas with significantly lower temperatures indicative of conditions such as peripheral artery disease, diabetic neuropathy, or hypothyroidism. Temperatures below 28°C are flagged as low. This can indicate conditions typically associated with reduced blood flow.

[0076] High Temperature

[0077]

[0047] Regions with elevated temperatures often associated with infections, inflammatory conditions, trauma, or injury. Temperatures above 34°C are flagged as high, which often suggests inflammation, infection, or trauma.

[0048] Additionally, asymmetrical temperature differences exceeding ±2.2°C (4°F) between corresponding regions of contralateral feet may be significant indicators of high-risk conditions such as developing ulcers or infections

[0078] Relative Deviations

[0079]

[0049] Relative deviations compare a specific region’s temperature to the average or surrounding temperatures within the same foot. This approach ensures that patientspecific variability is accounted for, improving diagnostic precision. As such, the step of applying the thermal images to the pre-trained classifier may comprise calculating a mean temperature of the foot and comparing each pixel’s temperature to the mean to identify deviations exceeding a predetermined relative threshold.

[0080]

[0050] Alternatively or additionally, the step of applying the thermal images to the pre-trained classifier may comprise calculating a mean temperature of a portion of the foot surrounding a pixel and comparing the pixel’s temperature to the mean to identify deviations exceeding a predetermined threshold. As an example, the predetermined relative threshold may be approximately ±2°C.

[0081]

[0051] The method may, in some embodiments, also include the step of comparing corresponding regions of the thermal image of contralateral feet to identify asymmetrical temperature differences exceeding a predefined contralateral threshold. Preferably, the predefined contralateral threshold is approximately ±2.2°C.

[0082]

[0052] The method, at box 225, may include the additional step of adjusting the classification of the region of the thermal image identified as exceeding the predetermined relative threshold and / or the predetermined contralateral threshold. The adjustment of the classification may be based on a direction and magnitude of the deviation.

[0053] As an example, the mean temperature of the entire foot or specific regions may be calculated. This may exclude areas affected by artifacts or extreme outliers. Each pixel’s temperature is compared to this mean and assessed for deviations.

[0083]

[0054] A deviation of more than approximately ±2°C from the mean temperature of the surrounding area or contralateral foot is flagged as abnormal (i.e., low or high classification).

[0084]

[0055] As set out above, the system initially classifies the temperature using the standardised thresholds:

[0085] ■ Normal: 28°C to 34°C

[0086] ■ Low: <28°C

[0087] ■ High: >34°C

[0088]

[0056] Regions within the normal range (28°C-34°C) are further analysed for deviations from the surrounding mean temperature. For example, a region with a temperature of 29°C (within the normal threshold) may show a significant deviation if the mean surrounding temperature is 33°C. As a result, embodiments of the method and the system may re-categorise that region initially categorised as normal as abnormal (low temperature).

[0089]

[0057] Conversely, a region at 35°C might be initially flagged as high, but if the surrounding mean is 34.8°C, the deviation is minimal, and the system adjusts the classification accordingly.

[0090]

[0058] In a clinical example, a patient undergoes thermal imaging of their feet, where the mean temperature of the right foot is determined to be 32°C based on the thermal images. A localised region on the thermal image of the right foot shows a temperature of 36°C, while the corresponding region on the thermal image of the contralateral left foot measures 31.5°C. Using the method described herein, thelocalised region is classified as high temperature according to the absolute threshold, as its temperature exceeded 34°C. Further evaluation of the thermal images according to the method reveals that the region's temperature deviates by +4°C from the mean temperature of the right foot and by +4.5°C from the corresponding region on the left foot. These deviations confirm the classification of the region as abnormal (high temperature). As a result, this region flagged for further clinical attention based on the thermal imaging analysis.

[0091]

[0059] This combination of standardized thresholds and relative deviations can be important for enhancing diagnostic outcomes in detecting foot abnormalities. Standardized thresholds establish universal baselines for temperature classifications, ensuring consistency across patients, while the inclusion of relative deviations enables personalized and patient-specific evaluations. This dual approach significantly improves diagnostic precision by accounting for individual variations in temperature profiles.

[0092]

[0060] Additionally, the integration of contextual analysis reduces the likelihood of false positives and negatives. By adjusting classifications to account for isolated temperature outliers, the system minimizes errors, providing more reliable diagnostic results. Furthermore, this method identifies even subtle changes in temperature, relative to the patient's baseline. These minor deviations, which might otherwise go unnoticed, can signal early-stage complications, enabling timely interventions that can prevent the progression of conditions such as diabetic foot ulcers. This combination ensures both accuracy and early detection, ultimately improving patient outcomes.

[0093]

[0061] Once the thermal images have been classified, in another step, at box 220, the thermal images are processed to generate segmented images by isolatingportions of the thermal images corresponding to the regions of the foot categorised as low or high temperature regions. That is, once an image is classified as abnormal (low or high temperature), the method includes segmenting the affected region within the thermal image.

[0094]

[0062] As an example, the Segformer-encoder Unet++ model may be used to precisely segment the affected region within the thermal image. The model or classifier is trained to perform edge detection of the abnormal regions and using the edges to segment those regions from within the image.

[0095]

[0063] Thus, the segmented portion may be automatically calculated and saved for comparison across visits, enabling progress tracking overtime.

[0096]

[0064] This quantitative tracking over time may be useful for follow-ups, as it allows clinicians to objectively measure ulcer progression or healing, unlike traditional subjective assessments.

[0097]

[0065] Examples of the thermal images being classified and segmented can be seen in Figures 4-6.

[0098]

[0066] Following segmentation of the thermal images, at box 230, the area of each region of the foot categorised as low or high temperature regions is calculated.

[0099]

[0067] In particular, the dimensions of the categorised regions of the thermal images are calculated and this data may be saved for clinical consultation and comparison with future scans.

[0100]

[0068] In another step, illustrated at box 235, the results of the segmentation (i.e., the segmented images) may be output to a display controlled by the processor. The original thermal images and / or an image of the foot may be displayed alongside the segmented images. In some embodiments, the segmented images may be overlaid on an image of the foot to show and highlight the regions of interest.

[0069] In some embodiments, another step, illustrated at box 240, includes receiving patient health metrics or data and processing the patient health metrics or data using a machine learning algorithm (e.g., XGBoost, Random Forest, Support Vector Machines) in combination with classifications of the thermal images of the foot to provide the risk assessment for DFU development. The patient health metrics or data may include blood pressure and / or glucose levels.

[0101]

[0070] Additionally, in some embodiments, the patient health metrics or data may include, but are not limited to one or more of the following: age, weight, height, smoking history, diet, physical activity level, family history of diseases, fasting blood glucose levels, random blood glucose levels, glycosylated haemoglobin (HbA1c), blood glucose levels, ankle-brachial index, urine test and / or blood pressure.

[0102]

[0071] Embodiments of the method and system may integrate these patient health metrics or data to enhance predictive accuracy, using ensemble learning models that assess DFU risk levels, thereby helping clinicians determine the likelihood of DFU development.

[0103]

[0072] In some embodiments, the integration of thermal imaging analysis with patient health metrics or data enhances diagnostic precision by correlating localized physiological abnormalities with broader systemic risk factors. Embodiments of the method involve processing thermal images of a foot to detect temperature anomalies and classify affected regions based on standardized thresholds and relative deviations, as previously described. Concurrently, patient-specific health data, including demographics, medical history, blood glucose levels, cholesterol, and blood pressure, may be analyzed using machine learning models, such as XGBoost, Random Forest, or logistic regression, as examples, to assist in the prediction the likelihood of DFU development. The method then fuses these outputs at a decisionlevel, using ensemble techniques such as weighted averaging or majority voting, to generate a comprehensive risk profile. For example, if thermal imaging identifies inflammation in a localized foot region and the patient’s health metrics indicate poor glucose control and high cholesterol, the system may assign a high DFU risk classification, prompting earlier intervention.

[0104]

[0073] The method may further include generating a report that integrates segmented thermal images with quantified ulcer metrics and systemic risk indicators, allowing clinicians to assess progression trends and adjust treatment plans accordingly. By combining objective thermal imaging data with individualized patient health insights, the system supports proactive DFU management, improves early detection capabilities, and facilitates personalized treatment recommendations, ultimately reducing the likelihood of severe complications such as infections or amputations.

[0105]

[0074] By way of example, in an embodiment, the method receives thermal images of a patient’s foot and applies a pre-trained classifier to identify abnormal temperature regions. The analysis reveals a 3°C temperature deviation in the midfoot region, indicating localized inflammation. Patient health metrics or data are also processed using a machine learning model trained on clinical risk factors for DFU development. The patient’s data indicates poor glucose control (HbA1c: 9.0%), high cholesterol levels, and a history of peripheral neuropathy, all of which contribute to an elevated risk of ulcer formation. The method integrates or combines these findings with the classifications of the thermal images of the foot by applying a decision-level fusion technique, such as weighted averaging or rule-based classification, to determine the overall DFU risk level. In this case, the combined analysis classifies the patient as high risk for DFU progression, given the presenceof both localized thermal abnormalities and systemic health concerns. A report is generated, displaying the segmented thermal images highlighting the affected region, alongside quantitative ulcer progression metrics and a DFU risk classification. The report further includes personalized treatment recommendations, such as the need for more frequent foot monitoring, glucose management interventions, or referral for specialist evaluation. By integrating thermal imaging results with patient-specific health data, the method provides a data-driven risk assessment that enables clinicians to initiate timely interventions, track ulcer progression over multiple imaging sessions, and tailor treatment strategies based on both visual and systemic indicators.

[0106]

[0075] In some embodiments, the method may be used across multiple imaging procedures.

[0107]

[0076] In such embodiments, thermal images of a foot are provided from a first imaging procedure and a subsequent imaging procedure as set out at box 310. The subsequent imaging procedure occurs at a time after the first imaging procedure.

[0108]

[0077] In some embodiments, the method may include the step of box 305, which includes performing imaging procedures to capture thermal images of a foot in a first imaging procedure and a subsequent imaging procedure.

[0109]

[0078] The steps of the method as set out above may be carried out on the thermal images captured from the first imaging procedure and the subsequent imaging procedure, such that regions of the thermal images from each imaging procedure are classified into categories based on temperature (see box 315), wherein the categories include low, normal, and high temperatures. The thermal images are subsequently segmented.

[0079] Embodiments of the method then involve processing the thermal images to generate segmented images at box 320 and comparing the segmented images from the first imaging procedure and the subsequent imaging procedure to identify changes in the regions classified as low or high temperature at box 325. The changes may take the form of changes in size / area and / or temperature across the imaging procedures.

[0110]

[0080] At box 330, the method may include calculating quantitative changes in the images. This may include the entirety of the foot to identify all changes. In some embodiments, calculating the quantitative changes includes calculating quantitative changes in the regions classified as low or high temperature.

[0111]

[0081] The method, at box 335, and system can then be used to generate a report and / or display indicating the changes identified between the imaging procedures to assist in treatment planning.

[0112]

[0082] Such embodiments performed across multiple imaging procedures may also incorporate patient health metrics or data as described elsewhere.

[0113]

[0083] Embodiments of the method described herein provide support for clinicians in the detection and management of diabetic foot ulcers (DFUs). Embodiments of the method provide early detection and precise localisation of DFUs, enabling timely interventions that reduce the risk of complications. Embodiments of the method generate quantitative data to track changes over time, allowing clinicians to monitor the progression or healing of abnormalities with objective metrics. Furthermore, embodiments can be used to produce comprehensive reports that integrate thermal imaging results with patient health metrics or data, offering a holistic assessment that aids in clinical decision-making.

[0084] The process employed by some embodiments of the method provide numerous advantages. For example, the method enables early detection by identifying subtle temperature deviations, even at Stage 0, before visible ulcers develop. This capability allows interventions to occur at the earliest and most effective stage, significantly reducing the risk of severe complications. The use of Al-powered segmentation and automated area calculations ensures that the data provided is objective, repeatable, and free from the subjective variability that often accompanies traditional methods. Additionally, embodiments provide integration of imaging, patient data, and follow-up tools into a single platform, enhancing the efficiency and effectiveness of clinical workflows. Its portability and non-contact design make it accessible for use in diverse settings, including underserved regions, thereby broadening its impact compared to traditional diagnostic methods, which are often resource-intensive and limited in application.

[0114]

[0085] These differences may have substantial implications. Early detection reduces the likelihood of severe complications, such as infections and amputations, improving patients' quality of life. The provision of quantitative metrics and Al-driven analysis enhances diagnostic precision, allowing clinicians to develop data-driven treatment plans. Finally, the seamless integration of patient data and imaging fosters personalized care, enabling clinicians to deliver timely and tailored interventions that address the specific needs of each patient. Collectively, these advancements represent a significant step forward in the management of DFUs.

[0115]

[0086] In compliance with the statute, the invention has been described in language more or less specific to structural or methodical features. The term “comprises” and its variations, such as “comprising” and “comprised of” is used throughout in an inclusive sense and not to the exclusion of any additional features.

[0087] It is to be understood that the invention is not limited to specific features shown or described since the means herein described comprises preferred forms of putting the invention into effect.

[0116]

[0088] The invention is, therefore, claimed in any of its forms or modifications within the proper scope of the appended claims appropriately interpreted by those skilled in the art.

Claims

CLAIMS1. A method executed by a computer processor for detecting abnormalities in a foot, the method comprising:receiving thermal images of a foot;applying the thermal images to a pre-trained classifier to classify regions of the foot into categories based on temperature, wherein the categories are one of: normal, low or high temperature; andprocessing the thermal images to generate segmented images by isolating portions of the thermal images corresponding to the regions of the foot categorised as low or high temperature; andcalculating an area of each region of the foot categorised as low or high temperature.

2. The method of claim 1, wherein the abnormal category comprises two subcategories: low temperature and high temperature.

3. The method of any one of the preceding claims, wherein the classifier further classifies regions classified as abnormal as one of: low or high temperature.

4. The method of any one of the preceding claims, wherein the method comprises receiving thermal images of two feet, including a left foot and a right foot of a patient.

5. The method of any one of the preceding claims, wherein the method comprises the step of capturing thermal images of a foot or feet.

6. The method of claim 3, wherein the regions are defined by a collection of pixels in the thermal images.

7. The method of any one of the preceding claims, wherein the step of applying the thermal images to the pre-trained classifier comprises applying predetermined temperature thresholds (standardised temperature thresholds) to classify regions of the foot, wherein temperatures within the range of 28°C to 34°C are classified as normal, temperatures below 28°C are classified as low, and temperatures above 34°C are classified as high.

8. The method of any one of claims 1 to 6, wherein the step of applying the thermal images to the pre-trained classifier comprises calculating a mean temperature of the foot and comparing each pixel’s temperature to the mean to identify deviations exceeding a predetermined relative threshold (relative temperature thresholds).

9. The method of claim 8, wherein the step of applying the thermal images to the pre-trained classifier comprises calculating a mean temperature of a portion of the foot surrounding a pixel and comparing the pixel’s temperature to the mean to identify deviations exceeding a predetermined threshold.

10. The method of claim 9, wherein the predetermined relative threshold is approximately ±2°C.

11. The method of any one of the preceding claims, wherein the method comprises the step of comparing corresponding regions of the thermal image of contralateral feet to identify asymmetrical temperature differences exceeding a predefined contralateral threshold.

12. The method of claim 11, wherein the predefined contralateral threshold is approximately ±2.2°C.

13. The method of any one of the preceding claims, wherein the method comprises the step of adjusting the classification of one or more regions of the thermal image.

14. The method of any one of the preceding claims, wherein the method comprises the step of adjusting the classification of one or more regions of the thermal images identified as exceeding the predetermined relative threshold and / or the predetermined contralateral threshold.

15. The method of any one of the preceding claims, wherein the adjustment of the classification is based on a direction and magnitude of the deviation.

16. The method of any one of the preceding claims, wherein the segmented images include only the regions of the foot categorised as low or high temperature regions.

17. The method of any one of the preceding claims, wherein the computing device is configured to perform the step of displaying the segmented images and the thermal images.

18. The method of any one of the preceding claims, wherein the method comprises the step of receiving patient health data and processing the patient health data using a machine learning algorithm in combination with classifications of the thermal images of the foot to provide a risk assessment for DFU development.

19. A system for processing a thermal image to detecting abnormalities in a foot, the system comprising:a computing device configured to perform the steps of:receiving thermal images of a foot;applying the thermal images to a pre-trained classifier to classify regions of the foot into categories based on temperature, wherein the categories are one of: normal, low or high temperature; andprocessing the thermal images to generate segmented images by isolating portions of the thermal images corresponding to the regions of the foot categorised as low or high temperature; andcalculating an area of each region of the foot categorised as low or high temperature.

20. A method executed by a computer processor for monitoring abnormalities in a foot over time, the method comprising:receiving thermal images of a foot from a first imaging procedure and a subsequent imaging procedure;applying the thermal images from each imaging procedure to a pre-trained classifier to classify regions of the foot into categories based on temperature, wherein the categories are one of: low, normal, and high temperatures;generating segmented images for each imaging procedure by isolating portions of the thermal images corresponding to regions of the foot classified as low or high temperature;comparing the segmented images from the first imaging procedure and the subsequent imaging procedure to identify changes between the thermal images from the first imaging procedure and thermal images from the subsequent imaging procedure; andgenerating a report indicating the changes identified between the imaging procedures to assist in treatment planning.

21. A method executed by a computer processor for detecting abnormalities in a foot, the method comprising:receiving thermal images of a foot;applying the thermal images to a pre-trained classifier to classify regions of the foot into categories based on temperature, wherein the categories are one of: normal and abnormal; andprocessing the thermal images to generate segmented images by isolating portions of the thermal images corresponding to the regions of the foot categorised as abnormal regions; andcalculating an area of each region of the foot categorised as abnormal regions.

Images