Spectral Rectification for Parameter-Efficient Adaptation of Foundation Models in Colonoscopy Depth Estimation

TL;DR

This paper introduces SpecDepth, a spectral rectification method that efficiently adapts foundation models for colonoscopy depth estimation by amplifying high-frequency components, leading to state-of-the-art results.

Contribution

The paper proposes a novel spectral rectification framework with a learnable wavelet module for domain adaptation in medical imaging, preserving geometric features while improving depth estimation.

Findings

Achieved state-of-the-art performance on C3VD and SimCol3D datasets.

Effectively amplifies high-frequency features to address spectral mismatch.

Demonstrated the importance of spectral alignment in model adaptation.

Abstract

Accurate monocular depth estimation is critical in colonoscopy for lesion localization and navigation. Foundation models trained on natural images fail to generalize directly to colonoscopy. We identify the core issue not as a semantic gap, but as a statistical shift in the frequency domain: colonoscopy images lack the strong high-frequency edge and texture gradients that these models rely on for geometric reasoning. To address this, we propose SpecDepth, a parameter-efficient adaptation framework that preserves the robust geometric representations of the pre-trained models while adapting to the colonoscopy domain. Its key innovation is an adaptive spectral rectification module, which uses a learnable wavelet decomposition to explicitly model and amplify the attenuated high-frequency components in feature maps. Different from conventional fine-tuning that risks distorting high-level semantic features, this targeted, low-level adjustment realigns the input signal with the original inductive bias of the foundational model. On the public C3VD and SimCol3D datasets, SpecDepth achieved state-of-the-art performance with an absolute relative error of 0.022 and 0.027, respectively. Our work demonstrates that directly addressing spectral mismatches is a highly effective strategy for adapting vision foundation models to specialized medical imaging tasks. The code will be released publicly after the manuscript is accepted for publication.