Virtual cutaneous area severity index (vCASI): A comprehensive methodology for quantitative skin disease assessment on positron emission tomography images.

purposeQuantification of skin diseases such as cutaneous lymphoma or psoriasis is important for pretreatment planning and response assessment. Currently, this quantification relies on clinical assessments, which are cumbersome, error-prone, and subject to inter-reader variability. FDG-PET/CT is a widely used molecular imaging technique for non-invasive detection and quantification of metabolically active disorders, providing biomarkers of disease extent, severity, and therapeutic response. However, skin disease quantification from PET/CT images remains challenging, error-prone, and labor-intensive.

methodsWe proposed a novel comprehensive methodology called virtual Cutaneous Area Severity Index (vCASI) to quantitatively assess the extent and severity of metabolically active skin disease from FDG-PET/CT images. Firstly, we used an automated body region localization method, followed by a standardization technique for PET images to reduce interscan measurement variability. Then, we generated skin shell PET images and developed a standardized Maximum Intensity Projection (MIP) rendering technique to enable reproducible 3D visualization of the skin with the disease. Finally, we introduced and validated a new quantitative scoring system (vCASI score) to measure the extent and severity of skin disease via these renderings.

resultsCorrelations between vCASI scores and reference standard ground-truth assessments increased following the standardization of PET images, further improved with the use of adjusted mean of maximum percentile intensity values as the reference point for MIP rendering, and reached as high as 0.605. Additionally, correlations between repeated vCASI score assessments for the torso, thorax, abdomen, and pelvis exceeded 0.85, demonstrating high repeatability.

conclusionsThe vCASI methodology allows for accurate and reproducible quantitative assessment of the extent and severity of metabolically active skin diseases, such as cutaneous lymphoma and psoriasis, from FDG-PET/CT images. It addresses the challenges related to body region localization, non-standardness of PET images, robust visualization, and standardized display of PET images to detect and quantify skin disease. Additionally, it overcomes the lack of a practical and validated scoring system for quantifying skin disease on PET images. We demonstrated that the vCASI scoring system has high repeatability and good accuracy, and is optimized by the proposed methodology.