基於簡易解剖量測的內耳畸形檢測：模型方法

// Detection of Inner Ear Malformations Based on Simple Anatomical Measurements: A Model Approach

本研究開發利用臨床影像中耳蝸基底回的簡易測量法，區分正常與內耳畸形耳蝸｜透過分析111例畸形與141例正常耳蝸的三維CT側壁螺旋，建立含基底回長度、耳蝸高度及比例的邏輯迴歸模型，整體準確率達94%｜模型能以最高90%準確率區分不完全分隔型與耳蝸發育不良類型，未來可作為術前規劃輔助工具，並支持自動化畸形檢測發展。

// This study developed a method using simple measurements of the cochlear basal turn from clinical imaging to distinguish normal cochleae from various inner ear malformations. By analyzing three-dimensional lateral wall spirals from CT scans of 111 malformed and 141 normal cochleae, logistic regression models based on basal turn length, cochlear height, and ratios accurately classified normal versus malformed cochleae with 94% accuracy. The models also differentiated incomplete partition and cochlear hypoplasia types with up to 90% accuracy, though some rare subtypes and distinctions required further data. This approach offers a fast, interpretable tool to aid preoperative planning and could support future automated malformation detection.

Published 2026年7月1日