Deep learning-based fetoscopic mosaicking for field-of-view expansion

Abstract

Purpose: Fetoscopic laser photocoagulation is a minimally invasive surgical procedure used to treat twin-to-twin transfusion syndrome (TTTS), which involves localization and ablation of abnormal vascular connections on the placenta to regulate the blood flow in both fetuses. This procedure is particularly challenging due to the limited field of view, poor visibility, occasional bleeding, and poor image quality. Fetoscopic mosaicking can help in creating an image with the expanded field of view which could facilitate the clinicians during the TTTS procedure.

Methods: We propose a deep learning-based mosaicking framework for diverse fetoscopic videos captured from different settings such as simulation, phantoms, ex vivo, and in vivo environments. The proposed mosaicking framework extends an existing deep image homography model to handle video data by introducing the controlled data generation and consistent homography estimation modules. Training is performed on a small subset of fetoscopic images which are independent of the testing videos.

Results: We perform both quantitative and qualitative evaluations on 5 diverse fetoscopic videos (2400 frames) that captured different environments. To demonstrate the robustness of the proposed framework, a comparison is performed with the existing feature-based and deep image homography methods.

Conclusion: The proposed mosaicking framework outperformed existing methods and generated meaningful mosaic, while reducing the accumulated drift, even in the presence of visual challenges such as specular highlights, reflection, texture paucity, and low video resolution.

Keywords: Deep learning; Fetoscopy; Sequential mosaicking; Surgical vision; Twin-to-twin transfusion syndrome (TTTS).

Fig. 1

Pictorial illustration of the fetoscopic laser photocoagulation for the treatment of TTTS and 1-to-1 mapping of 4-pt and $3\times 3$ homography parameterizations

Fig. 2

DIH regression network with controlled data generation for training

Fig. 3

Overview of the proposed FVM framework

Fig. 4

Representative frames from the five videos under analysis

Fig. 5

a–d Visualization of mosaics for one circular loop (360 frames) of the SYN sequence. e–g Quantitative comparison of FEAT, DIH and FVM

Fig. 6

Quantitative comparison of FVM, DIH, and FEAT on the test videos

Fig. 7

Qualitative results of the proposed FVM