{"id":12600,"date":"2026-06-20T15:52:04","date_gmt":"2026-06-20T10:22:04","guid":{"rendered":"https:\/\/eng.jfn.ac.lk\/mlsp\/?page_id=12600"},"modified":"2026-06-20T16:31:32","modified_gmt":"2026-06-20T11:01:32","slug":"genai-semantic-communication","status":"publish","type":"page","link":"https:\/\/eng.jfn.ac.lk\/mlsp\/?page_id=12600","title":{"rendered":"A Generative AI Framework for End-to-End Semantic Communication"},"content":{"rendered":"
Research Project · MLSP<\/span><\/p>\n

A Generative AI Framework for End-to-End Semantic Communication<\/h1>\n

Efficient medical image & video transmission through semantic communication<\/p>\n<\/div>\n

\n
\n

Overview<\/h2>\n

The rapid growth of wireless healthcare systems has increased the need to transmit medical images and videos efficiently and reliably. Traditional systems reproduce raw data bit-for-bit, but in medicine what matters most is preserving the clinically meaningful information<\/strong>. This project proposes a Generative-AI-based semantic communication framework<\/strong> that extracts, compresses, transmits and reconstructs the semantic content of medical images and videos instead of the full raw data — reducing bandwidth, improving robustness to channel noise, and preserving diagnostic relevance.<\/p>\n

Semantic features (anatomical structures, abnormal regions, disease-related patterns) are extracted with Vision Transformers<\/strong> and Large Language Models<\/strong>, compressed by a hybrid (classical + deep-learning) compressor, and — after a noisy wireless channel — reconstructed using CLIP-guided prompt embedding<\/strong> and a diffusion model<\/strong>, maintaining semantic integrity and perceptual quality.<\/p>\n<\/section>\n

\n

Methodology & Workflow<\/h2>\n

The framework moves compact semantic representations<\/em> across the channel instead of raw pixels — from transmitter, through the noisy channel, to generative reconstruction at the receiver.<\/p>\n

\n
\n
1<\/div>\n

Medical Input<\/h4>\n

Image \/ Video \/ Text<\/span><\/div>\n

\n
2<\/div>\n

Semantic Feature Extraction<\/h4>\n

ViT + LLM<\/span><\/div>\n

\n
3<\/div>\n

Encoding & Compression<\/h4>\n

Hybrid compressor<\/span><\/div>\n

\n
4<\/div>\n

Noisy Wireless Channel<\/h4>\n

Interference & distortion<\/span><\/div>\n

\n
5<\/div>\n

CLIP + Prompt → Diffusion<\/h4>\n

Guided reconstruction<\/span><\/div>\n

\n
6<\/div>\n

Semantic Reconstruction<\/h4>\n

Diagnostic output<\/span><\/div>\n<\/div>\n<\/section>\n

\n

Research Team<\/h2>\n
\n
\"Sandaruwan<\/p>\n

Sandaruwan W.P.P.C.<\/h4>\n
Undergraduate Researcher<\/div>\n
MLSP Research Group<\/div>\n<\/div>\n
\"Supunthaka<\/p>\n

Supunthaka W.H.S.S.<\/h4>\n
Undergraduate Researcher<\/div>\n
MLSP Research Group<\/div>\n<\/div>\n<\/div>\n<\/section>\n
← Back to Projects<\/a><\/div>\n<\/div>\n","protected":false},"excerpt":{"rendered":"

Research Project · MLSP A Generative AI Framework for End-to-End Semantic Communication Efficient medical image & video transmission through semantic communication Overview The rapid growth of wireless healthcare systems has increased the need to transmit medical images and videos efficiently and reliably. Traditional systems reproduce raw data bit-for-bit, but in medicine what matters most is preserving the clinically meaningful information. This project proposes a Generative-AI-based semantic communication framework that extracts, compresses, transmits and reconstructs the semantic content of medical images and videos instead of the full raw data — reducing bandwidth, improving robustness to channel noise, and preserving diagnostic relevance. Semantic features (anatomical structures, abnormal regions, disease-related patterns) are extracted with Vision Transformers and Large Language Models, compressed by a hybrid (classical + deep-learning) compressor, and — after a noisy wireless channel — reconstructed using CLIP-guided prompt embedding and a diffusion model, maintaining semantic integrity and perceptual quality. Methodology & Workflow The framework moves compact semantic representations across the channel instead of raw pixels — from transmitter, through the noisy channel, to generative reconstruction at the receiver. 1 Medical Input Image \/ Video \/ Text 2 Semantic Feature Extraction ViT + LLM 3 Encoding & Compression Hybrid compressor 4 Noisy Wireless Channel Interference & distortion 5 CLIP + Prompt → Diffusion Guided reconstruction 6 Semantic Reconstruction Diagnostic output Research Team Sandaruwan W.P.P.C. Undergraduate Researcher MLSP Research Group Supunthaka W.H.S.S. Undergraduate Researcher MLSP Research Group ← Back to Projects<\/p>\n","protected":false},"author":1,"featured_media":0,"parent":0,"menu_order":0,"comment_status":"closed","ping_status":"closed","template":"","meta":[],"_links":{"self":[{"href":"https:\/\/eng.jfn.ac.lk\/mlsp\/index.php?rest_route=\/wp\/v2\/pages\/12600"}],"collection":[{"href":"https:\/\/eng.jfn.ac.lk\/mlsp\/index.php?rest_route=\/wp\/v2\/pages"}],"about":[{"href":"https:\/\/eng.jfn.ac.lk\/mlsp\/index.php?rest_route=\/wp\/v2\/types\/page"}],"author":[{"embeddable":true,"href":"https:\/\/eng.jfn.ac.lk\/mlsp\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/eng.jfn.ac.lk\/mlsp\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=12600"}],"version-history":[{"count":1,"href":"https:\/\/eng.jfn.ac.lk\/mlsp\/index.php?rest_route=\/wp\/v2\/pages\/12600\/revisions"}],"predecessor-version":[{"id":12613,"href":"https:\/\/eng.jfn.ac.lk\/mlsp\/index.php?rest_route=\/wp\/v2\/pages\/12600\/revisions\/12613"}],"wp:attachment":[{"href":"https:\/\/eng.jfn.ac.lk\/mlsp\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=12600"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}