# Artificial Intelligence‐Enhanced Quantitative 3D Analysis of Distal Radioulnar Ligament Insertion Footprints of the Triangular Fibrocartilage Complex With Interactive Validation

## TL;DR

This study uses AI-enhanced MRI to accurately measure the 3D anatomy of wrist ligaments, improving clinical understanding of their structure.

## Contribution

A novel AI pipeline for MRI super-resolution and segmentation is validated against direct and 3D scanning measurements of DRUL footprints.

## Key findings

- AI-enhanced MRI showed excellent agreement with 3D scans for measuring DRUL footprint areas (ICC = 0.95-0.96).
- The most common shape of the deep DRUL footprint was a ribbon (64% of cases).
- AI methods demonstrated high reliability and accuracy compared to traditional anatomical measurements.

## Abstract

The distal radioulnar ligaments (DRULs) serve as primary stabilizers to the distal radioulnar joint (DRUJ). Existing cadaveric studies report heterogeneous morphometric data of the three‐dimensional (3D) anatomy of the triangular fibrocartilage complex (TFCC) and the ulnar footprints of the DRULs due to methodological variations and small sample sizes, limiting the translation of precise anatomical knowledge to clinical practice. This study quantitatively evaluated the 3D anatomy of the TFCC and the insertions of both superficial and deep DRULs components using three different methods with subsequent interactive validation: (1) direct measurement, (2) 3D scan, and (3) artificial intelligence (AI) enhanced magnetic resonance imaging.

Eleven adult cadaveric upper limbs were included. All specimens underwent 3.0‐Tesla MRI scans, which were then processed by AI algorithms for super‐resolution enhancement and semi‐automatic segmentation. The areas of deep and superficial limbs of DRUL ulnar footprint were measured in the super‐resolution MRI images using the Slicer software. The specimens were then dissected and anatomical measurements of dorsal‐volar maximal length and radial‐ulnar maximum length of deep ulnar DRUL footprint were performed on the specimens' photographs. Anatomical measurements of ulna, radius, triangular fibrocartilage, and ulnar insertions footprint of both superficial and deep DRULs were conducted subsequently using a 3D scanner. Primary outcome measures included the area and morphological classification (irregular quadrilateral, ribbon, semilunar) of the deep and superficial ulnar DRUL footprints. Statistical analysis encompassed intraclass correlation coefficients (ICC) for agreement assessment and multiple linear regression to explore associations.

The mean area of the deep foveal fibers of DRUL was 43.39 ± 13.49 mm2 and the superficial footprint was 20.11 ± 10.49 mm2 as measured with the 3D scanner. The morphologic features of the deep footprint shapes varied, with the most common shape being a ribbon (7/11, 64%). The intraclass correlation coefficients (ICCs) for the measurement of dorsal‐volar maximal length and radial‐ulnar maximum length of the DRUL between direct measurement and the 3D scan were excellent (ICC = 0.97 and 0.98, respectively). The ICCs between the AI‐enhanced analysis and the 3D scan for measuring the ulnar deep and superficial DRUL insertion areas were excellent (ICC = 0.95 and 0.96, respectively). Multiple linear regression explained 72.4% of the variance in deep DRUL footprint area (R
2 = 0.724, p = 0.147), with the superficial footprint area showing the strongest association (β = 0.639, p = 0.196).

Compared to direct measurement and 3D scan, the AI algorithms developed and validated for wrist MRI image enhancement demonstrated high accuracy and reliability in anatomical measurements of DRULs.

Interactive validation of a novel AI MRI super‐resolution and segmentation pipeline against direct anatomical measurement and 3D scanning for the precise quantification of distal radioulnar ligament (DRUL) insertion footprints.