Melanoma Diagnosis with Spatio-Temporal Feature Learning on Sequential Dermoscopic Images
This addresses misdiagnosis in borderline melanoma cases for dermatology by leveraging temporal changes, though it is incremental as it builds on existing image analysis with a new sequential approach.
The authors tackled automated melanoma diagnosis by using sequential dermoscopic images to capture lesion evolution over time, achieving an AUC of 74.34%, which is about 8% higher than single-image methods and 6% higher than LSTM-based models.
Existing studies for automated melanoma diagnosis are based on single-time point images of lesions. However, melanocytic lesions de facto are progressively evolving and, moreover, benign lesions can progress into malignant melanoma. Ignoring cross-time morphological changes of lesions thus may lead to misdiagnosis in borderline cases. Based on the fact that dermatologists diagnose ambiguous skin lesions by evaluating the dermoscopic changes over time via follow-up examination, in this study, we propose an automated framework for melanoma diagnosis using sequential dermoscopic images. To capture the spatio-temporal characterization of dermoscopic evolution, we construct our model in a two-stream network architecture which capable of simultaneously learning appearance representations of individual lesions while performing temporal reasoning on both raw pixels difference and abstract features difference. We collect 184 cases of serial dermoscopic image data, which consists of histologically confirmed 92 benign lesions and 92 melanoma lesions, to evaluate the effectiveness of the proposed method. Our model achieved AUC of 74.34%, which is ~8% higher than that of only using single images and ~6% higher than the widely used sequence learning model based on LSTM.