Acromegaly determination using discriminant analysis of the three-dimensional facial classification in Taiwanese

The aim of this study was to assess the size, angles and positional characteristics of facial anthropometry between "acromegalic" patients and control subjects. We also identify possible facial soft tissue measurements for generating discriminant functions toward acromegaly determination in males and females for acromegaly early self-awareness.

This is a cross-sectional study. Subjects participating in this study included 70 patients diagnosed with acromegaly (35 females and 35 males) and 140 gender-matched control individuals. Three-dimensional facial images were collected via a camera system. Thirteen landmarks were selected. Eleven measurements from the three categories were selected and applied, including five frontal widths, three lateral depths and three lateral angular measurements. Descriptive analyses were conducted using means and standard deviations for each measurement. Univariate and multivariate discriminant function analyses were applied in order to calculate the accuracy of acromegaly detection.

Patients with acromegaly exhibit soft-tissue facial enlargement and hypertrophy. Frontal widths as well as lateral depth and angle of facial changes were evident. The average accuracies of all functions for female patient detection ranged from 80.0-91.40%. The average accuracies of all functions for male patient detection were from 81.0-94.30%. The greatest anomaly observed was evidenced in the lateral angles, with greater enlargement of "nasofrontal" angles for females and greater "mentolabial" angles for males. Additionally, shapes of the lateral angles showed changes. The majority of the facial measurements proved dynamic for acromegaly patients; however, it is problematic to detect the disease with progressive body anthropometric changes.

The discriminant functions of detection developed in this study could help patients, their families, medical practitioners and others to identify and track progressive facial change patterns before the possible patients go to the hospital, especially the lateral "angles" which can be calculated by relative point-to-point changes derived from 2D lateral imagery without the 3D anthropometric measurements. This study tries to provide a novel and easy method to detect acromegaly when the patients start to have awareness of abnormal appearance because of facial measurement changes, and it also suggests that undiagnosed patients be urged to go to the hospital as soon as possible for acromegaly early diagnosis.