A new study suggests a novel deep-learning model based on diagnostic data may help harness the power of neural networks and artificial intelligence (AI) to more accurately diagnose melanoma.

"Biopsy is the traditional method for melanoma diagnosis, but this method lacks reliability," the paper authors wrote in Skin Research and Technology. "Therefore, it is important to apply new methods to diagnose melanoma effectively."

According to the study, the researchers sought "to develop a novel multi-modal deep learning method that successfully integrates advanced AI techniques with a comprehensive array of diagnostic data," adding that "the proposed research aims to transform the field of melanoma diagnostics by providing a more concerted and informative procedure." To achieve this, the research team constructed a dataset that included dermoscopic images, histopathological slides, and genomic profiles. They also developed a custom framework that used Convolutional Neural Networks (CNNs) for image data analysis and Graph Neural Networks (GNNs) for genomic data analysis. The researchers trained and evaluated the framework on this dataset.

According to the results, the multi-modal DNN had greater accuracy (92.5% and an area under the receiver operating characteristic curve of 0.96) compared with traditional medical approaches. These results indicate the model's ability to detect critical morphologic and molecular features of melanoma, surpassing the limitations of conventional AI and machine learning methods. 

"The combination of cutting-edge AI may allow access to a broader range of diagnostic data, which can allow dermatologists to make more accurate decisions and refine treatment strategies. However, the application of the framework will have to be validated at a larger scale and more clinical trials need to be conducted to establish whether this novel diagnostic approach will be more effective and feasible.

Source: Kiran A, et al. Skin Research and Technology. 2024. Doi:10.1111/srt.13770