A deep learning-integrated multimodal diagnostic framework for enhanced early detection of neuroendocrine tumours

#4512

Introduction: Neuroendocrine tumours (NETs) are a heterogeneous group of malignant neoplasms originating from neuroendocrine cells, commonly found in the digestive system, lungs, pancreas, and other organs. Due to their diverse clinical presentations and overlap with common diseases, the early diagnosis of NETs is challenging, often leading to misdiagnosis or delayed diagnosis at advanced stages. While current imaging techniques and biomarkers are useful in certain cases, their diagnostic sensitivity and specificity remain limited when used in isolation.

Aim(s): In this study, we propose a multimodal diagnostic algorithm that integrates clinical features and CT imaging characteristics to improve the early detection rate of NETs.

Materials and methods: Based on nearly 10 years of NET case data collected from the Second Affiliated Hospital of Kunming Medical University, including a total of 700 cases, we utilised deep learning algorithms to analyse CT images and clinical data for accurate identification of early NET lesions.

Conference:

Presenting Author:

Authors: Ding Z, Li H, Tang Z, Chen Y, Li H,

Keywords: Neuroendocrine Tumour, Multimodal Diagnostic algorithm, Deep Learning, Early Disease Screening,