Palm Image Analysis for Breast Cancer Risk Assessment: A Pilot Study on ATD Angle Asymmetry Using Machine Learning (Preprint)

BACKGROUND Breast cancer remains one of the leading causes of cancer-related mortality among women globally, with early detection playing a critical role in improving survival rates. However, traditional diagnostic methods such as mammography and ultrasound may be invasive, costly, or culturally stigmatized, leading to delayed diagnosis in certain populations. Palm print analysis has been proposed as a potential non-invasive biomarker in various medical conditions, but its role in breast cancer screening remains underexplored. OBJECTIVE This study aimed to investigate the feasibility of using palm image angle features—specifically atd, dat, and tda angles—as potential non-invasive indicators of breast cancer. It also sought to determine whether significant asymmetries exist between the left and right palms in patients with breast cancer compared to healthy controls. METHODS A total of 50 female participants were recruited and divided equally into a breast cancer group (n=25) and a healthy control group (n=25). Standardized palm images were captured and annotated using LabelMe, and angle measurements were performed using digital tools. The atd, dat, and tda angles were extracted from both hands using a Python-based image analysis workflow. Statistical analyses included independent-sample t-tests and paired-sample t-tests to compare angle differences between groups and within individuals. RESULTS Between-group comparisons revealed no statistically significant differences (p > .05) in atd, dat, or tda angles between the breast cancer and healthy groups. However, within the breast cancer group, significant asymmetries were found: the left-hand atd and tda angles were significantly larger than those of the right hand (p < .05), indicating potential palm print asymmetry in individuals with breast cancer. These findings align partially with prior literature and suggest a physiological relevance of palm angle asymmetry in cancer patients. CONCLUSIONS Although between-group differences were not significant, the detection of left–right palm asymmetry in breast cancer patients highlights a novel physiological trait that may contribute to early risk assessment. The use of palm image analysis presents a promising, non-invasive, and culturally sensitive screening tool, especially when integrated with machine learning algorithms. Future work should expand the sample size, include diverse ethnicities, and leverage AI techniques such as convolutional neural networks to enhance predictive accuracy. This research contributes to the growing body of digital health tools aimed at early detection and personalized health monitoring.