INTRODUCTION:

The skin prick test (SPT) stands as a cornerstone in the diagnosis of allergic reactions, offering valuable insights into individual sensitivities towards various allergens. This diagnostic procedure entails the application of allergen extracts to the skin, followed by controlled pricking to induce a localized allergic response, visually manifested as a wheal. The pivotal aspect of this test lies in the accurate measurement of wheal size, as it directly correlates with the degree of sensitization to specific allergens.
However, despite its clinical significance, the evaluation of SPT results has long been plagued by challenges stemming from manual measurement techniques. Conventional approaches, reliant on visual assessment and manual measurements using millimeter rulers, are susceptible to inherent limitations. Intra- and inter-observer variability, compounded by irregular wheal shapes and variations in skin pigmentation, often compromise the reliability and reproducibility of results. Such limitations not only hinder accurate diagnosis but also impede comparative analysis across different studies, thereby undermining the standardization and reliability of SPT outcomes.
To address these pressing concerns, recent advancements in medical imaging and machine learning have spurred the development of automated systems for SPT evaluation. These innovative approaches leverage cutting-edge technologies, including 3D imaging systems and sophisticated algorithms, to enhance the accuracy and efficiency of wheal measurement. By employing techniques such as pyramidal surface decomposition, Principal Component Analysis (PCA), and deep learning architectures like U-Net, these systems aim to automate the detection and quantification of wheal dimensions, thereby mitigating the impact of manual errors and subjectivity.
Moreover, the quest for standardized methodologies in SPT evaluation has gained momentum, driven by the imperative to minimize error sources and ensure consistency in diagnostic outcomes. Collaborative efforts, spearheaded by organizations such as the Global Allergy and Asthma European Network, have underscored the importance of adherence to standardized protocols, delineating key parameters such as the spacing between pricks and the threshold for positive reactions.
Against this backdrop, the present study endeavors to contribute to this burgeoning field by developing a robust model for automated measurement of SPT wheals. By integrating state-of-the-art imaging technologies with advanced machine learning algorithms, we seek to enhance the reliability, reproducibility, and standardization of SPT evaluations. Through this interdisciplinary approach, we envisage a paradigm shift in allergy diagnosis, facilitating more accurate clinical assessments, seamless data exchange, and comparative analysis of SPT results across diverse patient cohorts and research studies.

ABSTRACT:

Skin prick tests (SPTs) are pivotal in diagnosing allergic reactions, relying on the measurement of wheal size post-allergen exposure. However, manual measurement techniques are fraught with limitations, including intra- and inter-observer variability. This paper presents advancements in automated SPT evaluation, leveraging 3D imaging and machine learning. We discuss challenges in SPT assessment, emphasizing the need for standardized protocols. Our proposed model integrates pyramidal surface decomposition and PCA for wheal detection, utilizing a U-Net architecture for segmentation. By automating wheal measurement, our model aims to enhance reliability and standardization in allergy diagnosis, fostering improved clinical outcomes and research comparability.

PROBLEM STATEMENT:

Manual measurement techniques in skin prick test (SPT) evaluation suffer from intra- and inter-observer variability, hindering the reliability and standardization of allergy diagnosis. Existing automated methods often require extensive parameter tuning and may not be universally applicable. Furthermore, variations in wheal morphology and skin pigmentation pose additional challenges to accurate measurement. Consequently, there is a pressing need for a robust and universally applicable automated system for SPT wheal measurement, capable of overcoming the limitations of manual techniques and existing automated methods. This system should ensure reproducibility, standardization, and reliability in SPT evaluation, thereby facilitating more accurate allergy diagnosis and research outcomes.

OBJECTIVE:

The primary objective of this study is to develop an automated system for skin prick test (SPT) wheal measurement, addressing the inherent limitations of manual techniques and existing automated methods. Leveraging advancements in 3D imaging and machine learning, our system aims to achieve accurate and reproducible wheal measurements across diverse patient populations. To accomplish this, we propose the integration of pyramidal surface decomposition and Principal Component Analysis (PCA) for precise wheal detection, ensuring robust performance even in cases of irregular wheal morphology. Furthermore, employing a U-Net architecture enables us to achieve reliable segmentation of SPT wheals, overcoming challenges posed by variations in skin pigmentation and wheal appearance. A key focus of this endeavor is to streamline the automated system, minimizing the need for extensive parameter tuning and ensuring its universal applicability. Through rigorous validation against manual measurements and existing automated methods, we seek to demonstrate the superior performance of our system in terms of reproducibility, standardization, and reliability of SPT wheal measurements. Ultimately, the deployment of this automated system promises to enhance clinical diagnosis and research comparability in the field of allergy testing, facilitating more accurate assessments and data exchange across diverse healthcare settings.