In this issue, Mori et al. reported that a brand-new AI-CADx system (EndoBRAIN-Plus,
OLYMPUS) using endocytoscopy showed sufficient diagnostic capability in the discrimination
of invasive cancer and its high specificity may contribute to a reduction in unnecessary
surgeries for large adenomas. I have read this article with great interest. The history
of endocytoscopy and CADx began with a pilot trial conducted on the esophagus by Kodashima
et al [1]. Thereafter, there have been many reports concerning the usefulness thereof in the
identification of tumors and invasion depth diagnosis when applied in colon [2]
[3]
[4]
[5]. We have great interest in a part of this paper by Mori, which indicated that it
is possible to use AI diagnosis to check for the presence of invasive cancer in colorectal
lesions 2 cm or larger that are susceptible to overtreatment. However, they wrote
that, “Apparently, identification of the histologically worst area in a lesion is
a prerequisite of using EndoBRAIN-Plus in an appropriate way. We did not assess how
endoscopists identify these areas in the study, which will be an important research
topic in the future.” In other words, when actual EndoBRAIN-Plus-CADx is assumed,
it is necessary to identify the area of interest of the deepest part of the cancer
infiltration in the lesion. Therefore, endoscopists need to have sufficient understanding
of the fundamental diagnosis of colorectal lesions along with the methods of approaching
them, with beginners left with the question of how this is possible. This approach
suggests that, for the time being, it is realistic to apply CADx as an educational
training tool to support beginners in getting better at fundamental diagnostic endoscopy.
Also, because CADx cannot show the processes that led to a diagnosis (black box diagnosis),
research should also be conducted into whether it is possible to actually provide
this kind of education to beginners in a logical manner.
The majority of original papers on the subject of endocytoscopy come out of Japan.
There are few reports from countries other than Japan, so this area has not yet become
pervasive [6]
[7]. However, endocytoscopy can be sufficiently applied in everyday clinical applications.
Our hospital has been using it for everyday clinical applications since 2016. We confirmed
insertion time for the endocytoscope reaching the cecum was acceptable compared with
others (EC 5.5 min vs. Others 6.4 min). Because the magnification power of endocytoscopy
ranges from standard to ultra-high, we can obtain a standard image with NBI and the
cellular image with dye staining at the same time ([Fig. 1]). However, because endoscopic diagnosis with standard magnification also provides
sufficient diagnostic accuracy, there are very few cases in which routine clinical
applications require ultra-high magnification. The appeal of endocytoscopy is that
it is well designed for CADx, given its ability to easily yield uniform images with
a contact approach to colorectal lesions. Naturally, another appealing point is its
ability to observe cellular-level information in real time. Dissemination of CADx
before sufficient research has been performed into the clinical effectiveness of ultra-high
magnification, which is the original objective of endocytoscopy, could reduce the
enthusiasm that younger researchers have for diagnostic endoscopy with ultra-high
magnification. Shouldn’t we not only focus on what is available now, but also have
more discussions about the significance and appeal of ultra-high magnification endocytoscopy
in the future? We believe that surely, there exists a variety of yet-unknown information
still to be discovered.
Fig. 1 Simultaneous observation of a colon polyp (IIa + IIc, 8 mm, T1b adenocarcinoma) under
WLI, NBI, and chromoendoscopy using endocytoscopy. a WLI. b NBI (JNET Type 3). c Chromoendoscopy. d Cellular image (EC 3b, × 520).
