Abstract:While performing a colonoscopy for a 60-year-old gentleman on a cozy afternoon in November, you may find a relatively small adenoma in the sigmoid colon and remove it. If you estimate its size to be 10 mm, he may be back for surveillance colonoscopy in 3 years. If you estimate its size to be 9 mm, he may even be exempt from undergoing surveillance colonoscopy [1]. This makes a big difference in terms of the burden for patients, physicians, and healthcare systems. Furthermore, inconsistency in size estimation may affect treatment methods. Therefore, to deliver trustworthy endoscopy practice, we need to be very meticulous in polyp size estimation. The year 2023 has been a remarkable year for polyp size measurement, with the emergence of new technologies that can possibly help endoscopists estimate polyp size in a more accurate way than previously. Many previous studies have however suggested that we have very limited capability when it comes to size estimation for colorectal polyps, with accuracies varying from 54% to 65% [2] [3]. The use of a biopsy forceps as a measurement reference has been considered a classical solution, but it has not been widely accepted owing to its cumbersomeness, additional cost, and even unignorable inter-rater variation [4]. Do we have any solutions to overcome this clinically relevant issue? The answer to this question would have been "No" 10 years ago, but now we may humbly say "Yes" to this question in 2023. The year 2023 has been a remarkable year for polyp size measurement, with the emergence of new technologies that can possibly help endoscopists estimate polyp size in a more accurate way than previously. These innovative technologies include both hardware- and software-based approaches [5] [6]. Recently, Fujifilm Corporation provided a laser-based measurement device on the market. With the use of a special endoscope integrated with a laser function (SCALE EYE; Fujifilm Corp., Tokyo), the endoscopy processor lays a ruler over the endoscopic image. This ruler is automatically deployed onto the mucosa with reference to the distance between the tip of the endoscope and the mucosal surface where the laser light is shed on. This ruler-based measurement outperformed biopsy forceps-based measurement in a recently published preclinical trial [5]. Another exciting innovation was recently published by a German research group, who exploited the potential of artificial intelligence (AI) to estimate polyp size. The researchers came up with the idea of using the diameter of the auxiliary waterjet as a reference standard to accurately measure the polyp size on the screen with the aid of AI. This study deserved attention because the authors successfully eliminated the need to prepare a mechanical device for size measurement. However, regardless of whether it is a laser beam or waterjet, providing a reference standard has been considered mandatory for accurate size measurement. The biggest question so far is whether it is possible to omit such a reference standard in size measurement during colonoscopy. This has long been considered a challenge in technology [7]; however, in this issue of Endoscopy , Honggang Yu and colleagues introduce an astonishing AI tool with convincing data. Their proposed AI tool "ENDOANGEL-CPS" does not require the provision of a reference standard during the estimation of polyp size [8]. Nevertheless, the AI tool provided an amazingly high performance in this benchmark study; the relative accuracy of the AI model was significantly higher than that of endoscopists (89.9% vs. 54.7%; P <0.001). The trick behind this is that the authors used the reference standard in the training phase rather than in the validation phase. In other words, the authors developed measurable virtual colon models and used them to train the AI architecture. More specifically, they prepared three-dimensional (3D) computed tomography (CT) colonography for a total of 12 patients to develop the AI model. In these 3D CT colonographies, every single pixel was measurable because the absolute distance was derived from the CT images. By implementing a virtual endoscopy in these virtual 3D colonographies, the authors successfully taught the AI model how what is shown on the monitor corresponds to the actual size. Thanks to this unique training phase, endoscopists do not have to take care of the reference standard when using this AI model in clinical practice. I would pay considerable respect to this innovative work, not only because of the brilliant idea, but also because of the tremendous efforts to bring this from a scratch-level idea to implementation. Of course, this may be a first step before a potential big leap. Innovation needs to be validated in rigorously designed clinical trials. The authors plan to conduct a large-scale -Abstract Truncated-

Polyp Size Estimation by Generalizing Metric Depth Estimation and Monocular 3D Reconstruction

A Three-Dimensional Measurement Method for Binocular Endoscopes Based on Deep Learning

A real-time deep learning-based system for colorectal polyp size estimation by white-light endoscopy: development and multicenter prospective validation

Self-supervised monocular depth estimation for gastrointestinal endoscopy

An Ensemble Framework of Deep Neural Networks for Colorectal Polyp Classification

A geometry-aware deep network for depth estimation in monocular endoscopy

Endoscopic measurement of the size of gastrointestinal polyps using an electromagnetic tracking system and computer vision-based algorithm

Consisaug: A Consistency-based Augmentation for Polyp Detection in Endoscopy Image Analysis

Enhancing endoscopic measurement: validating a quantitative method for polyp size and location estimation in upper gastrointestinal endoscopy

New horizons in polyp size estimation

Establishing ground truth of polyp size, morphology, and volume using three-dimensional scanning

D2polyp-Net: A Cross-Modal Space-Guided Network for Real-Time Colorectal Polyp Detection and Diagnosis

Assessing generalisability of deep learning-based polyp detection and segmentation methods through a computer vision challenge

Sparse-to-dense coarse-to-fine depth estimation for colonoscopy

Colorectal Polyp Detection in Real-world Scenario: Design and Experiment Study

Photometric single-view dense 3D reconstruction in endoscopy

MCE-Net: polyp segmentation with multiple branch series-parallel attention and channel interaction via edge distribution guidance

MonoLoT: Self-Supervised Monocular Depth Estimation in Low-Texture Scenes for Automatic Robotic Endoscopy

Enhanced Scale-aware Depth Estimation for Monocular Endoscopic Scenes with Geometric Modeling

Quantitative Radiology: Automated Measurement of Polyp Volume in Computed Tomography Colonography Using Hessian Matrixbased Shape Extraction and Volume Growing