Tumor surgery faces challenges such as unclear lesion boundaries, complex anatomy, and intraoperative tissue deformation. Traditional navigation systems struggle with soft‑tissue dynamics and real‑time accuracy. A new review published in Intelligent Oncology examines how AI and multimodal imaging are driving next‑generation surgical navigation toward greater precision and safety.
Three integrated systems are identified:
Intelligent image analysis system – converts raw multimodal images (CT, MRI, PET) into clear 3D lesion models using deep learning‑based segmentation and fusion techniques.
Personalized surgical planning system – performs simulated resections, path design, and adaptive planning based on individual tumor biology and anatomical features.
Precision instrument execution system – enables robotic‑arm control, real‑time tool tracking, force feedback, and safety threshold management.
Despite rapid progress, the review identifies four key technical bottlenecks currently limiting clinical translation:
Accurate fusion and registration of multimodal heterogeneous images – cross‑modal registration (e.g., PET‑CT/MRI) is challenged by different data formats, patient motion, and soft‑tissue deformation during surgery. Real‑time elastic registration remains an open problem.
Intelligent tumor boundary recognition and safe margin planning – while U‑Net and nnU‑Net have advanced automatic segmentation, integrating complementary information from multiple imaging modalities for precise boundary delineation and margin planning is still difficult.
Dynamic tracking and safety control in complex surgical scenes – intraoperative tissue displacement, organ deformation, and instrument interactions require robust real‑time tracking and adaptive control algorithms.
Standardized data management and procedure optimization – full‑process management of navigation data (clinical, imaging, and patient information) is essential for consistent, reproducible surgical outcomes and continuous improvement.
Looking forward, the review outlines four intelligent development trends:
Real‑time closed‑loop multimodal perception–decision–execution – millisecond‑level fusion of imaging, physiological, and force‑tactile data, combined with reinforcement learning for dynamic replanning and risk warning.
Personalized planning via intelligent analysis of complex tumor imaging – AI‑driven boundary recognition, subclinical lesion detection, and augmented/virtual reality overlays for intuitive 3D spatial understanding.
Safe navigation execution through integration of perception and planning – human‑in‑the‑loop AI and real‑time decision support systems to enhance surgical safety.
Standardized surgical procedures via full‑process navigation data management – establishing data‑driven protocols, preoperative simulation, and iterative quality improvement mechanisms.
The authors conclude that the next breakthrough requires building trustworthy and interpretable human‑machine collaborative surgical intelligence. Explainable AI (e.g., attention mechanisms, concept bottleneck models) will be crucial to make AI decisions transparent and aligned with clinical reasoning. Overcoming these challenges will not only push the precision limits of tumor surgery but also establish new technological standards in intelligent oncology.
Full editorial available on ScienceDirect:
https://doi.org/10.1016/j.intonc.2026.100055
Contact Information for Intelligent Oncology:
LinkedIn: @IntelligentOncology
X: @IntelligentOnco
Facebook: @intelligentoncology
Email Address: editorialoffice@intelligent-oncology.net
Official Website: https://www.sciencedirect.com/journal/intelligent-oncology
Submission Link: https://www2.cloud.editorialmanager.com/intonc/default2.aspx
