3D reconstruction and artificial intelligence algorithms for the detection and tumor depth measurement of basal cell carcinoma in RCM-OCT images: A pilot study Conference Paper
| Authors: | Pan, A.; de Carvalho, N.; Silva, L.; Harris, U.; Dusza, S.; Sahu, A.; Kose, K.; Monnier, J.; Chen, C. S.; Jain, M. |
| Title: | 3D reconstruction and artificial intelligence algorithms for the detection and tumor depth measurement of basal cell carcinoma in RCM-OCT images: A pilot study |
| Conference Title: | Photonics in Dermatology and Plastic Surgery 2024 |
| Abstract: | The Reflectance Confocal Microscopy – Optical Coherence Tomography (RCM-OCT) device has demonstrated its effectiveness in the in vivo detection a nd depth a ssessment of basal cell ca rcinoma (BCC), though its interpretation can be cha llenging for novices. Artificia l intelligence (AI) has the potential to a ssist in identifying BCC and measuring its depth in these images. Our goa l wa s to develop an AI model capable of generating 3D volumetric representations of BCC to enhance its detection and depth measurement. We developed AI models tra ined on OCT images of biopsyconfirmed BCC to detect BCC, generate 3D volumetric representations, and automatically a ssess tumor depth. These models were then tested on a separate dataset containing images of B CC, benign lesions, a nd normal skin. The effectiveness of the AI models wa s evaluated through a blinded reader study a nd by comparing tumor depth mea surements with those obtained from histopathology. The a ddition of AI-generated 3D renders of BCC improved BCC detection ra tes, with sensitivity increasing from 73.3% to 86.7% a nd specificity from 45.5% to 48.5%. A Pea rson Correla tion coefficient r2 = 0.59 (p=0.02) wa s a chieved in comparing tumor depth measurements between AI -generated renders a nd histopathology slides. Incorporating AI-generated 3D renders has the potential to improve the diagnosis of BCC a nd the automated measurement of tumor depth in OCT ima ges, reducing rea der dependent variability and sta ndardizing dia gnostic a ccuracy. © 2024 SPIE. |
| Keywords: | reflectance confocal microscopy; confocal microscopy; skin cancer; skin; optical tomography; tumors; artificial intelligence; diagnosis; reflectance confocal microscopies; reflection; dermatology; image reconstruction; skin cancers; optical coherence tomography; three dimensional computer graphics; basal cells; rendering (computer graphics); 3d imaging; intelligence models; 3d imaging reconstruction; skin cancerpreventionand early detection; depth measurements; imaging reconstruction; measurements of |
| Journal Title | Progress in Biomedical Optics and Imaging - Proceedings of SPIE |
| Volume: | 12816 |
| Conference Dates: | 2024 Jan 27-28 |
| Conference Location: | San Francisco, CA |
| ISBN: | 1605-7422 |
| Publisher: | SPIE |
| Date Published: | 2024-01-01 |
| Start Page: | 128160B |
| Language: | English |
| DOI: | 10.1117/12.3008592 |
| PROVIDER: | scopus |
| DOI/URL: | |
| Notes: | Conference paper -- Source: Scopus |
Altmetric
Citation Impact
BMJ Impact Analytics
Related MSK Work