Description:[001] The present invention discloses the evaluation and assessment of facial asymmetry, particularly in individuals suffering from facial palsy or related neural disorders. Through the utilization of the MediaPipe framework, key facial landmarks are identified and analyzed to discern asymmetry or drooping characteristics, employing techniques such as Partial Curve Mapping (PCM) and other mathematical algorithms for comparison and quantification.
BACKGROUND OF THE INVENTION
[002] Facial asymmetry, often stemming from neural disorders such as facial palsy, has become a growing concern affecting individuals of all ages, regardless of their medical history. This condition arises from the inability to contract facial nerves, leading to drooping faces and deformation, which not only impacts physical appearance but also induces psychological distress, hindering social interactions and communication due to discomfort in expressing emotions. Recent studies have also linked facial deformation to early signs of strokes, underscoring the importance of early detection. Leveraging advancements in technology, particularly in computer vision methods, offers a promising avenue for accurately evaluating facial asymmetry. In this context, the proposed invention focuses on analyzing facial images to generate key facial points using the MediaPipe framework and employing techniques like Partial Curve Mapping (PCM) to assess asymmetry or drooping faces, thereby providing a quantitative measure for future diagnosis.
[003] Facial paralysis, often synonymous with facial nerve paralysis or facial palsy, is a condition caused by temporary or permanent damage to the facial nerve, resulting in impaired muscle contraction in the affected area of the face. This impairment not only affects the ability to express emotions but also poses challenges in everyday communication, such as difficulty in moving the eyes, lips, and conveying natural emotions. The severity of facial paralysis varies, ranging from affecting both sides of the face to specific portions. Various factors contribute to facial paralysis, including nerve injuries, viral infections like Ramsay Hunt Syndrome, traumas such as surgeries or accidents, and even childbirth. With the incidence of facial paralysis increasing, there is a pressing need for accurate diagnostic methods to assess the degree of asymmetry and quantify facial deformities.
[004] Related works in the field of computer vision and medical imaging have explored various approaches for identifying and diagnosing facial paralysis. Studies by Kostiantyn Khabarlak et al. and Gemma S. Parra-Dominguez have provided insights into conventional and deep learning-based methods for facial landmark identification, highlighting their applications and limitations. Samuel Susan Veeravalli et al. have proposed deep learning image classification techniques for identifying facial deformities, emphasizing the importance of early diagnosis. Similarly, Jiang Chaoqun et al. have examined current approaches for assessing facial paralysis, underscoring the need for more reliable and accurate techniques.
[005] Among the proposed methods, ensemble regression tree-based facial feature extraction by Jocelyn Barbosa et al. and hierarchical network-based techniques by Gee-Sern Jison Hsu et al. have shown promise in accurately classifying facial paralysis. Additionally, innovative telemedicine systems like the Tele Stroke System (TSS) by Chandaliya et al. have facilitated remote diagnosis and treatment, offering timely intervention for conditions like stroke. Furthermore, advancements in human pose estimation systems, such as Google's MediaPipe posture package, provide valuable tools for analyzing facial features and assessing asymmetry.
[006] Despite these advancements, existing techniques still have limitations in terms of accuracy, availability, and cost-effectiveness. The proposed invention aims to overcome these shortcomings by leveraging computer vision methods to provide a robust system for evaluating facial asymmetry, aiding in early diagnosis and treatment planning for patients with facial palsy and related conditions. By integrating techniques like key facial point generation and mathematical analysis, the invention offers a comprehensive solution for quantifying facial deformities and facilitating timely intervention to improve patient outcomes.
SUMMARY OF THE PRESENT INVENTION
[007] The present invention addresses the significant issue of facial asymmetry resulting from various neural disorders. Facial asymmetry not only affects the physical appearance of individuals but also leads to psychological distress, hindering their ability to express themselves effectively in social interactions. Moreover, it has been identified as an early sign of strokes, underscoring the importance of timely detection and assessment.
[008] To address these challenges, the proposed methodology leverages advancements in technology, particularly computer vision methods, to analyze facial images comprehensively. The key innovation lies in the utilization of the MediaPipe framework, specifically its "BlazeFace" model for facial landmark identification and "FaceMesh" pre-trained model for detecting facial landmarks. These models enable precise identification of key facial features, essential for evaluating asymmetry.
[009] The methodology involves several key steps. Firstly, a dataset of facial images, focusing on individuals with facial paralysis, is collected and annotated to mark key features and facial outlines. Subsequently, training data preparation involves pre-processing and training machine learning models to recognize and track facial key features and contours.
[010] The integration of MediaPipe into the research pipeline facilitates contour extraction and landmark recognition, enabling the comprehensive analysis of facial asymmetry. Through model training and validation, the efficacy of the proposed methodology is assessed, with performance metrics such as accuracy and mean squared error evaluated.
[011] The evaluation of asymmetry involves analyzing key facial points, particularly those around the eyes and jawline. By computing differences in areas between key points of the eyes and comparing the similarity of jawline curves on either side of the face, the degree of asymmetry is quantified.
[012] The results obtained from sample images demonstrate the effectiveness of the proposed methodology. Images with facial palsy exhibit significant differences in asymmetry metrics compared to those without palsy, validating the approach's ability to accurately assess facial asymmetry.
[013] In conclusion, the proposed invention offers a comprehensive and innovative solution for evaluating facial asymmetry in individuals with neural disorders. By leveraging computer vision techniques and advanced methodologies, it provides valuable insights for diagnosis, treatment, and monitoring of facial palsy, thereby enhancing patient care and outcomes. Furthermore, the generated key points can be utilized for analyzing additional parameters, contributing to a deeper understanding of facial palsy and its management.
BRIEF DESCRIPTION OF THE DRAWINGS
[014] when considering the following thorough explanation of the present invention, it will be easier to understand it and other objects than those mentioned above will become evident. Such description refers to the illustrations in the annex, wherein:
FIGS. 1-2, illustrates systematic diagrams related to a facial palsy image evaluation system, in accordance with an embodiment of the present invention.
DETAILED DESCRIPTION OF THE INVENTION
[015] The following sections of this article will provide various embodiments of the current invention with references to the accompanying drawings, whereby the reference numbers utilised in the picture correspond to like elements throughout the description. However, this invention is not limited to the embodiment described here and may be embodied in several other ways. Instead, the embodiment is included to ensure that this disclosure is extensive and complete and that individuals of ordinary skill in the art are properly informed of the extent of the invention.
[016] Numerical values and ranges are given for many parts of the implementations discussed in the following thorough discussion. These numbers and ranges are merely to be used as examples and are not meant to restrict the claims' applicability. A variety of materials are also recognised as fitting for certain aspects of the implementations. These materials should only be used as examples and are not meant to restrict the application of the innovation.
[017] Referring now to the drawings, these are illustrated in FIGS. 1-2,
the present invention addresses the pressing issue of facial asymmetry resulting from various neural disorders, particularly facial palsy. This condition not only leads to physical deformities but also impacts individuals psychologically, hindering their ability to express themselves and engage in social interactions effectively. Furthermore, early detection of facial deformation is crucial as it can serve as an indicator of underlying health issues such as strokes.
[018] The proposed methodology leverages advancements in computer vision technology to analyze facial images accurately. Key to this approach is the utilization of the MediaPipe framework, which facilitates the identification of facial landmarks and contours essential for evaluating facial asymmetry. Specifically, the BlazeFace model within MediaPipe is employed for precise facial landmark identification, followed by the utilization of the FaceMesh model to detect facial landmarks.
[019] The methodology involves several steps, beginning with data collection focused on individuals with facial paralysis, followed by annotation to mark key facial features and contours. Subsequently, training data preparation involves the creation of training and testing subsets within the annotated dataset for machine learning model training. MediaPipe integration is crucial for contour and facial key point detection, requiring the selection and integration of specific components into the research pipeline. The development of the pipeline follows a graph-based approach, coupling selected calculators to create a comprehensive data processing pipeline.
[020] Model training utilizes annotated data to train machine learning algorithms, potentially employing deep learning techniques for accurate facial feature recognition. Validation and evaluation of the pipeline's performance are conducted using the testing subset of the annotated dataset, employing metrics such as accuracy, mean squared error, and intersection over union. Finally, experimentation and analysis involve the application of the developed pipeline in real-world scenarios, including the evaluation of facial paralysis treatments and therapies.
[021] The evaluation of facial asymmetry encompasses various aspects, including the analysis of key points generated from facial images. Notably, differences in eye characteristics and jawline shapes are significant indicators of facial asymmetry. The methodology involves the calculation of differences in the area of the eyes and the comparison of jawline curves using techniques like Partial Curve Mapping (PCM).
[022] Results obtained from the methodology demonstrate its effectiveness in evaluating facial asymmetry. Sample images depicting faces with and without palsy are analyzed, showcasing the capability of the proposed approach to accurately identify and quantify facial deformities. The computed values for asymmetry parameters highlight significant differences between palsy-affected and healthy faces, validating the methodology's efficacy.
[023] In conclusion, the proposed invention offers a comprehensive solution for evaluating facial asymmetry, particularly in individuals with facial palsy. By leveraging computer vision technology and advanced methodologies, it provides clinicians and researchers with valuable insights for diagnosis, treatment evaluation, and prognosis assessment. Additionally, the generated facial key points hold potential for further analysis of various facial parameters, contributing to a deeper understanding of facial paralysis and its management.
, Claims:1. A facial palsy image evaluation system comprising:
a) an image acquisition module for acquiring facial images of individuals with facial palsy;
b) a facial feature detection module for detecting facial landmarks and extracting facial contours from the acquired images;
c) an asymmetry quantification module for analyzing the extracted facial landmarks and contours to quantify facial asymmetry; and
d) a diagnostic reporting module for generating diagnostic reports based on the asymmetry indices and quantitative measurements obtained from the analysis.
2. The system as claimed in claim 1, wherein the facial feature detection module utilizes computer vision algorithms for accurate detection of facial landmarks and contours.
3. The system as claimed in claim 1, wherein the asymmetry quantification module employs mathematical algorithms to calculate asymmetry indices and quantitative measurements.
4. The system as claimed in claim 1, wherein the diagnostic reporting module generates comprehensive reports providing insights into the severity of facial palsy and aiding in treatment planning and monitoring.