Leonardo Ayala, Tim J. Adler, Silvia Seidlitz et al.
Science Advances, Vol 9, Issue 10
10 March 2023
Introduction
(excerpt)
Replacing traditional open surgery with minimally invasive techniques for complicated interventions such as tumor resection is one of the most important challenges in modern health care. Conventionally used RGB (red, green, and blue) camera-based laparoscopes, however, are ill-suited for these demands. Their mode of operation is based on mimicking the human eye by collecting light in the aforementioned three broad regions of the optical spectrum; as a consequence, precise tissue differentiation and assessment of organ function remain largely inaccessible. Yet, obtaining real-time functional tissue information is crucial for many key procedures in minimally invasive surgery.
A common necessity of laparoscopic surgeries, for example, is stopping the blood flow to a specific organ or tissue region by clamping the arteries responsible for blood supply. This process, commonly referred to as ischemia induction, prevents excessive bleeding of patients (1) and is performed in various procedures, including partial nephrectomy, organ transplantation, and anastomosis. After clamping the main arteries, it is highly challenging to assess the perfusion state of the tissue solely based on the available RGB video stream. This especially holds true for selective clamping of a segmental artery, in which ischemia is induced only in the cancerous part of the kidney during partial nephrectomy (2, 3). The most common approach to ensure correct clamping is based on indocyanine green (ICG) fluorescence (Fig. 1): After ICG is injected into the bloodstream, it binds to plasma proteins. The bound ICG travels through the bloodstream and accumulates in the internal organs, especially in the kidney and liver, within a minute (4, 5). Lack of a fluorescent signal thus corresponds to lack of perfusion. However, because of long washout periods of about 30 min, this test is not easily repeatable if the wrong segment has been clamped (5) or if the clamping procedure was done improperly, as illustrated in Fig. 1. Furthermore, it requires a contrast agent to be injected into the bloodstream. Although ICG injection is generally regarded as a safe procedure, cases with severe complications such as anaphylactic shock have been observed (6).
Spectral imaging is a promising alternative approach to improving surgical vision (7). This technique removes the arbitrary restriction of recording only three broad spectral bands (RGB) by capturing an n-dimensional feature vector for each camera pixel, where each dimension corresponds to a comparatively narrower spectral band. As different tissue structures have unique optical scattering and absorption properties, knowledge of these optical properties along with spectral measurement data can potentially provide important information on tissue morphology (8–10) and function (11, 12). The term multispectral imaging (MSI) is used when only a few bands (up to dozens) are recorded, while hyperspectral imaging (HSI) refers to hundreds of bands being recorded (7).
Despite the general success of MSI and HSI (7, 13–24), applications in the operating room (OR) have been limited. Some of the main reasons why spectral imaging has not yet found its way into surgical practice are related to image acquisition time, processing time, and size of the available devices (7). Many available MSI/HSI camera systems are large (14 to 50 cm) and/or take several seconds (2 to 8 s) to record and process one image (24–28). To the best of our knowledge, the only laparoscopic spectral device proposed for clinical use so far (26) takes around 5 s to record one hyperspectral image, which prevents real-time application. In consequence, clinical success stories in spectral imaging for minimally invasive surgery are still lacking. Specifically, we are not aware of any clinical study in the broader context of real-time perfusion monitoring based on spectral imaging in laparoscopic surgery.
We address this gap in the literature with the following contributions:
1) Video-rate MSI system (Fig. 2): We present the real-time (25 Hz) laparoscopic MSI system applied in patient studies. It features a compact (26 mm by 26 mm by 31 mm) and lightweight (32 g) MSI camera that can be connected to standard laparoscopes via a widely used C-Mount adapter and operates with clinical light sources.
2) Deep learning (DL)–based algorithm for real-time ischemia detection (see Fig. 3): To overcome the need of large amounts of training data required by traditional discriminative machine learning methods, we phrase the problem of ischemia detection as an out-of-distribution (OoD) detection problem that does not rely on data from any other patient (patent pending). Using an ensemble of invertible neural networks (INNs) as a core component, our algorithm is trained to compute the likelihood of ischemia based on a short (several seconds) video sequence acquired at the beginning of each surgery.
3) In-human study: We present an in-human study demonstrating that monitoring kidney ischemia in real time is now possible.
Moves in Introdcutions (overview)
Establishling a territory: Why your project is important.
Establishing a niche: Why your article is necessary.
Occupying the niche: What you are going to do with this article.
Source: Macgilchrist (2014)
Moves in Introductions (details)
Comments