Publication

Multimodal Learning for Lung Cancer Diagnosis and Management: A Deep Learning Pipeline for Classification, TNM Staging, and Treatment Protocol Generation

View document

Bibliographic Details
Summary:Accurate lung cancer diagnosis and staging are critical for personalized treatment and improved patient outcomes. Traditional diagnostic methods, such as manual image examination, are prone to variability and inefficiency. Recent advancements in Deep Learning (DL) have demonstrated potential in automating lung cancer classification and staging, thereby enhancing diagnostic accuracy and efficiency. However, existing solutions often address only isolated aspects of the diagnostic process, such as tumor detection, without offering a unified system for multi-target classification that integrates clinical tools for both clinicians and patients. This study presents a unified framework for lung cancer analysis that combines medical imaging, clinical data, and Large Language Models (LLMs) to support three key tasks: tumor type classification, TNM staging, and automated treatment protocol recommendation. Image-based classification was performed using YOLOv8n, trained on two CT datasets, achieving a maximum mean average precision (mAP50) of 0.418 and an F1 score of 0.44. TNM staging was addressed through a multimodal classifier, combining the ResNet50 model with a multilayer perceptron, which fused imaging and demographic inputs. This approach yielded an average F1 score of 0.389, with the M component showing the strongest performance. For treatment generation, a Retrieval-Augmented Generation (RAG) approach was employed, combining clinical prompts with relevant documents to produce personalized protocols using the Gemini 2.0 Flash LLM. The best-performing configuration achieved a stage match of 0.54 and a BERTScore of 0.83, along with high contextual fidelity across RAGAS metrics (Faithfulness: 0.68, Answer Relevancy: 0.81, Context Precision: 0.99, and Context Recall: 0.93). This framework demonstrates the potential to support both diagnosis and treatment planning within a single integrated system, contributing to more personalized and effective clinical decision-making in oncology.
Main Authors:Silva, Catarina Costa Pereira Nascimento da
Subject:Deep Learning (DL) Large Language Models (LLMs) Lung Cancer Retrieval-Augmented Generation (RAG) Treatment Recommendation
Year:2025
Country:Portugal
Document type:master thesis
Access type:open access
Associated institution:Universidade Nova de Lisboa
Language:English
Origin:Repositório Institucional da UNL
_version_ 1868413809032429568
author Silva, Catarina Costa Pereira Nascimento da
author_facet Silva, Catarina Costa Pereira Nascimento da
author_role author
contributor_name_str_mv Castelli, Mauro
Jardim, João Bruno Morais de Sousa
RUN
country_str PT
creators_json_txt [{\"Person.name\":\"Silva, Catarina Costa Pereira Nascimento da\"}]
datacite.contributors.contributor.contributorName.fl_str_mv Castelli, Mauro
Jardim, João Bruno Morais de Sousa
RUN
datacite.creators.creator.creatorName.fl_str_mv Silva, Catarina Costa Pereira Nascimento da
datacite.date.Accepted.fl_str_mv 2025-10-28T00:00:00Z
datacite.date.available.fl_str_mv 2025-11-10T10:44:23Z
datacite.date.embargoed.fl_str_mv 2025-11-10T10:44:23Z
datacite.rights.fl_str_mv http://purl.org/coar/access_right/c_abf2
datacite.subjects.subject.fl_str_mv Deep Learning (DL)
Large Language Models (LLMs)
Lung Cancer
Retrieval-Augmented Generation (RAG)
Treatment Recommendation
datacite.titles.title.fl_str_mv Multimodal Learning for Lung Cancer Diagnosis and Management: A Deep Learning Pipeline for Classification, TNM Staging, and Treatment Protocol Generation
dc.contributor.none.fl_str_mv Castelli, Mauro
Jardim, João Bruno Morais de Sousa
RUN
dc.creator.none.fl_str_mv Silva, Catarina Costa Pereira Nascimento da
dc.date.Accepted.fl_str_mv 2025-10-28T00:00:00Z
dc.date.available.fl_str_mv 2025-11-10T10:44:23Z
dc.date.embargoed.fl_str_mv 2025-11-10T10:44:23Z
dc.format.none.fl_str_mv application/pdf
dc.identifier.none.fl_str_mv http://hdl.handle.net/10362/190384
dc.language.none.fl_str_mv eng
dc.rights.cclincense.fl_str_mv http://creativecommons.org/licenses/by/4.0/
dc.rights.none.fl_str_mv http://purl.org/coar/access_right/c_abf2
dc.subject.none.fl_str_mv Deep Learning (DL)
Large Language Models (LLMs)
Lung Cancer
Retrieval-Augmented Generation (RAG)
Treatment Recommendation
dc.title.fl_str_mv Multimodal Learning for Lung Cancer Diagnosis and Management: A Deep Learning Pipeline for Classification, TNM Staging, and Treatment Protocol Generation
dc.type.none.fl_str_mv http://purl.org/coar/resource_type/c_bdcc
description Accurate lung cancer diagnosis and staging are critical for personalized treatment and improved patient outcomes. Traditional diagnostic methods, such as manual image examination, are prone to variability and inefficiency. Recent advancements in Deep Learning (DL) have demonstrated potential in automating lung cancer classification and staging, thereby enhancing diagnostic accuracy and efficiency. However, existing solutions often address only isolated aspects of the diagnostic process, such as tumor detection, without offering a unified system for multi-target classification that integrates clinical tools for both clinicians and patients. This study presents a unified framework for lung cancer analysis that combines medical imaging, clinical data, and Large Language Models (LLMs) to support three key tasks: tumor type classification, TNM staging, and automated treatment protocol recommendation. Image-based classification was performed using YOLOv8n, trained on two CT datasets, achieving a maximum mean average precision (mAP50) of 0.418 and an F1 score of 0.44. TNM staging was addressed through a multimodal classifier, combining the ResNet50 model with a multilayer perceptron, which fused imaging and demographic inputs. This approach yielded an average F1 score of 0.389, with the M component showing the strongest performance. For treatment generation, a Retrieval-Augmented Generation (RAG) approach was employed, combining clinical prompts with relevant documents to produce personalized protocols using the Gemini 2.0 Flash LLM. The best-performing configuration achieved a stage match of 0.54 and a BERTScore of 0.83, along with high contextual fidelity across RAGAS metrics (Faithfulness: 0.68, Answer Relevancy: 0.81, Context Precision: 0.99, and Context Recall: 0.93). This framework demonstrates the potential to support both diagnosis and treatment planning within a single integrated system, contributing to more personalized and effective clinical decision-making in oncology.
dirty 0
eu_rights_str_mv openAccess
format masterThesis
fulltext.url.fl_str_mv https://run.unl.pt/bitstreams/73dc5028-a35e-4245-90bf-6ada1d07e0d8/download
id run_6c877dfd4e71e776f486069fb7b12ffb
identifier.url.fl_str_mv http://hdl.handle.net/10362/190384
instacron_str unl
institution Universidade Nova de Lisboa
instname_str Universidade Nova de Lisboa
language eng
network_acronym_str run
network_name_str Repositório Institucional da UNL
oai_identifier_str oai:run.unl.pt:10362/190384
organization_str_mv urn:organizationAcronym:unl
person_str_mv Silva, Catarina Costa Pereira Nascimento da
publishDate 2025
reponame_str Repositório Institucional da UNL
repository_id_str urn:repositoryAcronym:run
service_str_mv urn:repositoryAcronym:run
spelling engpt_PTAccurate lung cancer diagnosis and staging are critical for personalized treatment and improved patient outcomes. Traditional diagnostic methods, such as manual image examination, are prone to variability and inefficiency. Recent advancements in Deep Learning (DL) have demonstrated potential in automating lung cancer classification and staging, thereby enhancing diagnostic accuracy and efficiency. However, existing solutions often address only isolated aspects of the diagnostic process, such as tumor detection, without offering a unified system for multi-target classification that integrates clinical tools for both clinicians and patients. This study presents a unified framework for lung cancer analysis that combines medical imaging, clinical data, and Large Language Models (LLMs) to support three key tasks: tumor type classification, TNM staging, and automated treatment protocol recommendation. Image-based classification was performed using YOLOv8n, trained on two CT datasets, achieving a maximum mean average precision (mAP50) of 0.418 and an F1 score of 0.44. TNM staging was addressed through a multimodal classifier, combining the ResNet50 model with a multilayer perceptron, which fused imaging and demographic inputs. This approach yielded an average F1 score of 0.389, with the M component showing the strongest performance. For treatment generation, a Retrieval-Augmented Generation (RAG) approach was employed, combining clinical prompts with relevant documents to produce personalized protocols using the Gemini 2.0 Flash LLM. The best-performing configuration achieved a stage match of 0.54 and a BERTScore of 0.83, along with high contextual fidelity across RAGAS metrics (Faithfulness: 0.68, Answer Relevancy: 0.81, Context Precision: 0.99, and Context Recall: 0.93). This framework demonstrates the potential to support both diagnosis and treatment planning within a single integrated system, contributing to more personalized and effective clinical decision-making in oncology.application/pdfpt_PTMultimodal Learning for Lung Cancer Diagnosis and Management: A Deep Learning Pipeline for Classification, TNM Staging, and Treatment Protocol GenerationSilva, Catarina Costa Pereira Nascimento daCastelli, MauroJardim, João Bruno Morais de SousaHostingInstitutionOrganizationalRUNe-mailmailto:run@unl.ptrun@unl.ptURNurn:tid:2040730222025-11-10T10:44:23Z2025-10-282025-10-28T00:00:00ZHandlehttp://hdl.handle.net/10362/190384http://purl.org/coar/access_right/c_abf2open accessDeep Learning (DL)Large Language Models (LLMs)Lung CancerRetrieval-Augmented Generation (RAG)Treatment Recommendation7581445 bytesliteraturehttp://purl.org/coar/resource_type/c_bdccmaster thesis2025-10-28http://creativecommons.org/licenses/by/4.0/http://purl.org/coar/access_right/c_abf2application/pdffulltexthttps://run.unl.pt/bitstreams/73dc5028-a35e-4245-90bf-6ada1d07e0d8/download
spellingShingle Multimodal Learning for Lung Cancer Diagnosis and Management: A Deep Learning Pipeline for Classification, TNM Staging, and Treatment Protocol Generation
Silva, Catarina Costa Pereira Nascimento da
Deep Learning (DL)
Large Language Models (LLMs)
Lung Cancer
Retrieval-Augmented Generation (RAG)
Treatment Recommendation
status SINGLETON
subject.fl_str_mv Deep Learning (DL)
Large Language Models (LLMs)
Lung Cancer
Retrieval-Augmented Generation (RAG)
Treatment Recommendation
title Multimodal Learning for Lung Cancer Diagnosis and Management: A Deep Learning Pipeline for Classification, TNM Staging, and Treatment Protocol Generation
title_full Multimodal Learning for Lung Cancer Diagnosis and Management: A Deep Learning Pipeline for Classification, TNM Staging, and Treatment Protocol Generation
title_fullStr Multimodal Learning for Lung Cancer Diagnosis and Management: A Deep Learning Pipeline for Classification, TNM Staging, and Treatment Protocol Generation
title_full_unstemmed Multimodal Learning for Lung Cancer Diagnosis and Management: A Deep Learning Pipeline for Classification, TNM Staging, and Treatment Protocol Generation
title_short Multimodal Learning for Lung Cancer Diagnosis and Management: A Deep Learning Pipeline for Classification, TNM Staging, and Treatment Protocol Generation
title_sort Multimodal Learning for Lung Cancer Diagnosis and Management: A Deep Learning Pipeline for Classification, TNM Staging, and Treatment Protocol Generation
topic Deep Learning (DL)
Large Language Models (LLMs)
Lung Cancer
Retrieval-Augmented Generation (RAG)
Treatment Recommendation
topic_facet Deep Learning (DL)
Large Language Models (LLMs)
Lung Cancer
Retrieval-Augmented Generation (RAG)
Treatment Recommendation
url http://hdl.handle.net/10362/190384
visible 1