基於人工智慧技術融合影像组學和臨床數據於腎細胞癌子型診斷

Abstract

本研究聚焦於腎細胞癌，特別是其主要類別腎亮細胞癌（Clear Cell Renal Cell Carcinoma, ccRCC），此為腎臟癌的主要發病原因。傳統上，醫生診斷腎細胞癌需要判斷腫瘤的良惡性及其具體類型，這通常依賴於醫療影像及組織切片等方法。然而，這些方法不僅耗時且有時結果不夠精確。本研究的重要性在於提出一種結合影像組學和臨床數據的新方法，以模擬醫生的診斷流程並進行改良，使用2021年腎腫瘤分割（KiTS21）公開資料集中的電腦斷層掃描（CT）影像，進行腎臟及腫瘤特徵的精確擷取。研究首先將進行腫瘤良惡性的分類，然後進一步區分ccRCC與非ccRCC腫瘤類型。透過這種兩階段分類法，本研究不僅旨在提升診斷的準確性和速度，同時也期望減少依賴侵入性生物切片的需求，從而減輕病人的負擔。在應用場景方面，此技術可直接應用於臨床診斷，協助醫生迅速且準確地判斷腎腫瘤的類型，進而制定更適合的治療方案。此外，這一技術的發展和完善也將推動個人化醫療的進程，使醫療資源能更有效地被利用。綜合上述，本研究的成果將對腎臟癌的診斷和治療產生深遠的影響，為醫療領域帶來顯著的進步。

This study zeroes in on Renal Cell Carcinoma (RCC), with a specific focus on its primary subtype, Clear Cell Renal Cell Carcinoma (ccRCC), a leading cause of kidney cancer. Traditionally, physicians diagnose RCC by determining the tumor's malignancy and specific type, often relying on medical imaging and tissue biopsy. However, these methods are not only time-consuming but sometimes yield less accurate results. The significance of this research lies in its proposition of a novel method that amalgamates radiomics with clinical data to refine and simulate the physician's diagnostic process. Utilizing Computed Tomography (CT) images from 2021 Kidney Tumor Segmentation datasets (KiTS21), the study accurately extracts kidney and tumor characteristics. It begins with classifying tumors into benign or malignant categories, followed by further differentiation into ccRCC and non-ccRCC types. Through this two-tier classification method, the study aims to enhance the accuracy and speed of diagnosis while also hoping to reduce the dependency on invasive biopsies, thereby alleviating patient distress. In practical scenarios, this technique can be directly applied to clinical diagnostics, aiding physicians in swiftly and precisely determining the type of renal tumor, thereby crafting more suitable treatment strategies. Moreover, the development and refinement of this technology are set to propel the progress of personalized medicine, ensuring more effective utilization of medical resources. In sum, the outcomes of this research are poised to profoundly impact the diagnosis and treatment of kidney cancer, marking a significant stride forward in the medical field.