CLASSICAL ENGINEERING VS. DEEP SPATIO-TEMPORAL PARADIGMS IN VIDEO POLYP SEGMENTATION: A SYSTEMATIC COMPARATIVE REVIEW

Badalova Lobar Burhonovna

Innores · 2026-yil

Annotatsiya

This systematic review compares three foundational paradigms inVideo Polyp Segmentation (VPS) for automated colonoscopy: classical engineering,static 2D deep learning, and recurrent spatio-temporal architectures. While the fieldhas shifted from geometric rules to data-driven networks, balancing frame-levelaccuracy with temporal consistency remains a critical engineering challenge. Usingthe multi-center SUN-SEG database, we establish a structural taxonomy byevaluating each paradigm's mathematical formulation, failure modes, and throughputunder clinical artifacts like specular reflections, motion blur, and out-of-view events.Our synthesis reveals that classical hand-crafted methods offer deterministicexplainability but fail under imaging noise. Memory-less 2D deep networks achievehigh spatial accuracy but suffer from boundary flickering and tracking dropouts dueto an inter-frame blind spot. Conversely, recurrent spatio-temporal hybrids exhibitsuperior resilience; by integrating bottleneck gating mechanics, they leveragehistorical hidden states to stabilize boundaries and project polyp shapes throughintense noise without sacrificing real-time throughput. This mapping outlines keyarchitectural trade-offs, serving as a deployment reference for future video-streamintelligence frameworks.

Maqola ma’lumotlari
MualliflarBadalova Lobar Burhonovna
JurnalInnores
Nashr sanasi2026-06-30
Jild2
Son6
Betlar173-188
TilIngliz

Kalit so‘zlar

Video Polyp Segmentation; Spatio-Temporal Modeling; ConvLSTM

Ilmiy soha

Innores jurnalidan boshqa maqolalar

Innores — barcha maqolalar