A Pathologically Validated B-Mode Ultrasound Classification System for Surgical Planning in Dorsocervical Fibrofatty...

Area of Science:

• Medical imaging
• Surgical pathology
• Biostatistics

Background:

• Dorsocervical fibrofatty hump (buffalo hump) lacks objective classification criteria.
• This leads to suboptimal surgical outcomes due to incorrect technique selection.

Purpose of the Study:

• Develop a data-driven algorithm for sub-clustering dorsocervical fibrofatty hump.
• Utilize histopathology, shear-wave elastography (SWE), and B-mode ultrasound (B-US) for classification.
• Facilitate improved therapeutic decision-making.

Main Methods:

• Analyzed 86 patients using Masson-stained histology and ImageJ for collagen quantification.
• Classified subtypes into adipose-dominant, mixed, and fibrous-dominant using K-means clustering.
• Determined diagnostic thresholds for SWE and B-US via decision tree analysis.

Main Results:

• Identified three distinct collagen subgroups with a Silhouette score of 0.661.
• B-mode ultrasound (B-US) demonstrated high accuracy (93.4%/92.0%) in classification.
• Shear-wave elastography (SWE) showed low accuracy (67.2%/48.0%) with overlapping classifications.

Conclusions:

• Collagen percentage provides a reliable classification for dorsocervical fibrofatty hump.
• B-mode ultrasound (B-US) is a dependable noninvasive tool for preoperative planning.
• Adipose-dominant lesions suit liposuction (SAL/UAL), while fibrous-dominant cases require excision.