[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-128779-en":3,"doc-seo-128779-105":30,"detail-sidebar-cat-0-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":4,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":27,"seo_description":14,"update_tm":28,"read_time":29},128779,1099523882367,"Hazel","https://ap-avatar.wpscdn.com/davatar_9964176cb1d06d4a9deccf72a44ae3dc",8,"Research & Report","Medical Image Segmentation with Selective State Space Models - Thesis","Medical image segmentation is a core task in clinical diagnostics, where precision and reliability directly affect downstream decisions. Manual delineation is labor intensive, motivating automated methods as reliable alternatives. While Transformer-based approaches have driven recent progress, their scalability and computational efficiency limitations encourage exploration of Selective State Space Models. This thesis evaluates Mamba-based architectures, U-Mamba and Swin-UMamba, on multiple polyp segmentation datasets using tailored augmentation to improve generalization and reduce overfitting.","DEPARTMENT OF INFORMATION ENGINEERING BACHELOR’S DEGREE IN COMPUTER ENGINEERING  \nMedical Image Segmentation with Selective State  \nSpace Models  \nThesis Supervisor  \nProf. Carlo Fantozzi  \nGraduating Student  \nAlberto Levorato  \nACADEMIC YEAR 2024-2025  \nGraduation Date 24/07/2025  \nTo my family and the friends that I met along the way  \nAbstract  \nMedical image segmentation is a critical task in clinical diagnostics, requiring high precision and reliability. Traditional manual segmentation is labor intensive, prompting the need for automated approaches. While Transformer-based architectures have dominated recent advances, their limitations in scalability and computational efficiency have motivated exploration of alternative models. This paper investigates Mamba, a novel class of Selective State Space Models, as a promising alternative for medical image segmentation. Two Mamba-based architectures U-Mamba and Swin-UMamba have been evaluated across multiple polyp segmentation datasets, with custom data augmentation techniques applied to improve generalization and mitigate overfitting. Results demonstrate that Mamba models offer competitive performance, with Swin-UMamba approaching state-of-the-art accuracy. Thanks to their efficiency, adaptability, and compact size, these models are well-suited for emerging medical applications. Overall, the findings suggest that Mamba architectures represent a promising direction for future research in medical computer vision.  \nSommario  \nLa segmentazione delle immagini mediche  un compito fondamentale nella diagnostica clinica, che richiede alta precisione e affidabilit . La segmentazione manuale  un processo lavorativamente intenso, rendendo necessarie soluzioni automatizzate. Sebbene i recenti progressi nel campo siano stati dominati da architetture basate su Transformer, le loro limitazioni in termini discalabilit ed efficienza computazionale hanno spinto alla ricerca di modelli alternativi. Questo studio analizza Mamba, una nuova classe di modelli Selective State Space, come alternativa promettente per la segmentazione di immagini mediche. Sono state valutate U-Mamba e SwinUMamba, due architetture Mamba, su diversi dataset di segmentazione di polipi, applicandotecniche dedicate di data augmentation per migliorare la generalizzazione e ridurre l’overfitting. I risultati dimostrano che i modelli Mamba offrono prestazioni competitive, con Swin-UMamba che si avvicina allo stato dell’arte. Grazie alla loro efficienza, adattabilit e dimensioni compatte, questi modelli risultano adatti a nuove applicazioni in ambito medico. I risultati suggeriscono che le architetture Mamba rappresentano una direzione promettente per la ricerca futura nella computer vision applicata alla medicina.  \nContents  \n1 Introduction 1  \n1.1 State Space model ................................. 2  \n1.2 Structured State Space method .......................... 4  \n1.3 Selective State Space model ........................... 4  \n2 Mamba Models 7  \n2. 1 U-Mamba . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8  \n2.2 Swin-UMamba . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8  \n3 Datasets 11  \n3. 1 AbdomenMR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11  \n3.2 Kvasir-SEG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12  \n3.3 HarDNet-MSEG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13  \n4 Experiments 14  \n4. 1 Installation and validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14  \n4.2 Training and evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15  \n4.2. 1 Kvasir-SEG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15  \n4.2.2 HarDNet-MSEG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16  \n4.3 Data augmentation ................................ 18  \n4.3.1 Data Augmentation 1 (DA1) ....................... 18  \n4.3.2 Data Augmentation 2 (DA2) ..","cbCainudlP5ERsUN","https://ap.wps.com/l/cbCainudlP5ERsUN","pdf",2203799,1,28,"English","en",105,"# Introduction\n## State Space model\n## Structured State Space method\n## Selective State Space model\n# Mamba Models\n## U-Mamba\n## Swin-UMamba\n# Datasets\n## AbdomenMR\n## Kvasir-SEG\n## HarDNet-MSEG\n# Experiments\n## Installation and validation\n## Training and evaluation\n## Data augmentation\n## Data Augmentation 1 (DA1)\n## Data Augmentation 2 (DA2)\n# Conclusions","[{\"question\":\"Why is automated medical image segmentation needed instead of manual segmentation?\",\"answer\":\"Manual segmentation is labor intensive and time-consuming. Automated approaches aim to provide high precision and reliability for clinical use.\"},{\"question\":\"What problem with Transformer-based architectures motivates alternative models?\",\"answer\":\"Transformers can have limitations in scalability and computational efficiency, especially due to attention complexity over finite windows.\"},{\"question\":\"Which Mamba-based architectures are evaluated for medical image segmentation, and what datasets are used?\",\"answer\":\"The thesis evaluates U-Mamba and Swin-UMamba on multiple polyp segmentation datasets, including AbdomenMR, Kvasir-SEG, and HarDNet-MSEG.\"}]","Medical Image Segmentation with Selective State Space Models - Thesis | PDF",1786003357,71,{"code":4,"msg":31,"data":32},"ok",{"site_id":24,"language":23,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":28},"medical-image-segmentation-with-selective-state-space-models-thesis","",{"@graph":36,"@context":85},[37,54,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,51],{"item":41,"name":42,"@type":43,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":50},"https://docshare.wps.com/document/research-report/",3,{"item":52,"name":13,"@type":43,"position":53},"https://docshare.wps.com/document/medical-image-segmentation-with-selective-state-space-models-thesis/128779/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":23,"description":14,"dateModified":62,"datePublished":62,"encodingFormat":61,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":41,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-08-06",true,{"@type":65,"interactionType":66,"userInteractionCount":4},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"Why is automated medical image segmentation needed instead of manual segmentation?","Question",{"text":75,"@type":76},"Manual segmentation is labor intensive and time-consuming. Automated approaches aim to provide high precision and reliability for clinical use.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"What problem with Transformer-based architectures motivates alternative models?",{"text":80,"@type":76},"Transformers can have limitations in scalability and computational efficiency, especially due to attention complexity over finite windows.",{"name":82,"@type":73,"acceptedAnswer":83},"Which Mamba-based architectures are evaluated for medical image segmentation, and what datasets are used?",{"text":84,"@type":76},"The thesis evaluates U-Mamba and Swin-UMamba on multiple polyp segmentation datasets, including AbdomenMR, Kvasir-SEG, and HarDNet-MSEG.","https://schema.org",{"og:url":52,"og:type":87,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":89,"canonical":52},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,101,105,110,115,120,123,128,131,135],{"id":20,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":46,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":46,"category_name":112,"show_sort_weight":113,"slug":114},6,"Technology",50,"technology",{"id":116,"doc_module":4,"doc_module_name":46,"category_name":117,"show_sort_weight":118,"slug":119},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":121,"slug":122},30,"research-report",{"id":124,"doc_module":4,"doc_module_name":46,"category_name":125,"show_sort_weight":126,"slug":127},9,"Religion & Spirituality",20,"religion-spirituality",{"id":126,"doc_module":4,"doc_module_name":46,"category_name":129,"show_sort_weight":126,"slug":130},"World Cup","world-cup",{"id":132,"doc_module":4,"doc_module_name":46,"category_name":133,"show_sort_weight":132,"slug":134},10,"Lifestyle","lifestyle",{"id":136,"doc_module":4,"doc_module_name":46,"category_name":137,"show_sort_weight":106,"slug":138},19,"General","general"]