Page 16 - Read Online

Page 261 Collins et al. Art Int Surg. 2025;5:254-69 https://dx.doi.org/10.20517/ais.2024.92

Enhanced surgery visualization and safety: AI systems can integrate data from multiple sources to provide
surgeons with enhanced visualization. For instance, we have developed AI models to automatically detect
and track seven key anatomical structures in transabdominal preperitoneal (TAPP) inguinal hernia
[16]
repair . This AI system, which can integrate with robot arm movement data, may be used to continually
visualize structures even if they are obscured by bleeding or a prosthetic. This visualization could enhance
safety in inguinal hernia surgery by reducing complications, especially chronic postoperative pain due to
improperly located prosthetic staples, and preventing vascular injury near the “Triangle of Doom”.

Procedure monitoring, standardization, and quality assurance: Procedure guidelines, such as the “10 Golden
Rules” in MIS inguinal hernia repair , were proposed to improve hernia repair quality and standardization.
[42]
However, they are not widely adopted due to a lack of awareness/training, clinical practice variability, and
[16]
guideline complexity. AI systems may significantly increase guideline adoption by continually monitoring
surgical video data and validating guideline conformity in real time.

We have made the first advances in this direction by developing an AI system that detects critical structures
that must be exposed to ensure a complete TAPP inguinal hernia dissection . This model may be
[16]
integrated with a mesh detection model to alert the surgeon if they attempt to place a mesh over an
inadequately dissected hernia. We have also developed an AI model to divide TAPP surgical videos into
seven main surgical steps automatically . The steps may be nonsequential, typical in bilateral repairs, and
[17]
the system can process live and archive videos. It has various applications, including quantifying time spent
on each step, which can be applied to enhance OR efficiency by, e.g., alerting the support staff to prepare the
mesh when the dissection is nearing completion. In the future, these AI models will be enhanced in analysis
depth, e.g., quantitative assessment of dissection zones, mesh coverage, and tissue tension, and their
application breadth for use in different abdominal wall procedures.

Automated surgery reports: AI systems are being developed to analyze surgical images and videos and to
[43]
generate automated reports . They are likely to be adapted to AWS soon. This may reduce the burden on
the surgical staff while also providing a comprehensive and systematic surgical record of every procedure.
Using video-based recognition models, they may also provide information beyond that of standard reports,
including a detailed log of events and surgical actions/decisions.

Emergency surgery
Abdominal wall emergencies, such as incarcerated or strangulated hernias, pose significant challenges that
differ from elective procedures. In these situations, time is critical, and surgeons must make rapid decisions
to prevent complications like bowel ischemia, perforation, and sepsis. Unlike planned surgeries, where there
is time for thorough preoperative imaging and detailed surgical planning, emergency cases often require
immediate action based on clinical judgment and limited diagnostic information.

One of the most pressing concerns in emergency AWS is determining whether bowel loops trapped within a
hernia sac remain viable or need resection. This decision is traditionally made based on visual inspection
and subjective assessment of factors like bowel color, peristalsis, and pulsation. However, misjudging bowel
viability can lead to either unnecessary resections or, conversely, leaving necrotic tissue in place, both of
which can result in severe complications. AI-assisted imaging techniques, such as real-time perfusion
analysis using near-infrared fluorescence [44,45] or hyperspectral imaging [46,47] , could help surgeons make more
accurate assessments and reduce errors in intraoperative decision making.