A new deep learning algorithm successfully distinguishes between desmoplastic and non-desmoplastic histopathological growth patterns in colorectal liver metastases, achieving high discriminatory power with area under the curve values of 0.93 and 0.95 during development and external validation, respectively. This automated classification parallels manual scoring regarding overall survival outcomes, ensuring its potential utility in routine histopathological assessments. The method leverages extensive digitalized imaging from a large patient cohort to enhance predictive accuracy for patient prognosis.
Journal Article by Höppener DJ, Aswolinskiy W (…) Verhoef C et 11 al. in BJS Open
© The Author(s) 2024. Published by Oxford University Press on behalf of BJS Foundation Ltd.