Depth-Dependent Corneal Backscatter Characteristics in Different Stages of Keratoconus and a Novel Composite...

To characterise the depth-dependent distribution of corneal backscatter across keratoconus stages and controls, and to develop a composite Corneal Backscatter Index (CBSI) integrating multi-depth features for early detection of forme fruste keratoconus (FFKC) METHODS: In this retrospective comparative study, 213 eyes of 213 participants were enrolled: 65 FFKC, 47 clinical keratoconus (CK), 30 severe keratoconus (SK), and 71 healthy controls. On Scheimpflug tomography images, the cornea at the thinnest point was divided into 50 layers, and backscatter greyscale values were extracted at 2% depth intervals. Receiver operating characteristic (ROC) analysis assessed the discriminative ability of each layer for FFKC. Forward stepwise logistic regression then combined layer backscatter with established tomographic parameters, including the Belin/Ambrósio deviation value (BAD-D), to derive the CBSI as a probability score.

Corneal backscatter alterations depended on depth and varied with keratoconus severity. Compared with controls, FFKC showed significantly increased backscatter within the 0%-8% and 24%-48% depth ranges (P < 0.05). Relative to FFKC, CK showed further increases at the anterior 12% depth and posterior cornea (52%-82%), with a decrease at 94%-100%. SK demonstrated higher backscatter than CK throughout the 12%-100% range. For FFKC identification, the CBSI, combining CBS_26%-28%, CBS_90%-92% and CBS_96%-98% with BAD-D, yielded an AUC of 0.943 (95% CI, 0.909-0.977), significantly outperforming BAD-D alone (AUC = 0.767).

Corneal backscatter varied significantly with both depth and keratoconus severity. In FFKC, depth‑dependent backscatter alterations were detectable before morphologic abnormalities, providing an early diagnostic signal. The CBSI improved FFKC diagnostic performance over morphologic parameters alone.