As vantagens clínicas do diagnóstico de cancro de próstata assistido por inteligência artificial: um estudo piloto inovador do Ipatimup

Um novo estudo feito no Ipatimup avaliou de forma independente a ferramenta de inteligência artificial (IA) Paige Prostate, aprovada pela FDA, na prática clínica. Os resultados, publicados na Pathology-Research and Practice, mostraram que os diagnósticos anatomo-patológicos assistidos por IA foram entregues 24 horas mais cedo, concordam melhor com os achados nas peças cirúrgicas, melhoraram a identificação de pacientes do Grupo Grau 1 em 27% dos casos e podem ajudar a evitar cirurgias desnecessárias em pacientes com classificação incorretamente superestimada. Este é um trabalho importante para o avanço da compreensão do papel da IA ​​na patologia.

 

Autores e afiliações:

C. Eloy ^a,b,*,1, A. Asaturova ^c,1, J. Pinto ^b, I. Rienda ^b, A. Syrnioti ^d, R. Prisco ^e, A. Polonia ^b,f 

^a Pathology Department, Medical Faculty of University of Porto, Porto, Portugal 

^b Pathology Laboratory, Institute of Molecular Pathology and Immunology of University of Porto (IPATIMUP), Porto, Portugal 

^c 1st pathology department FSBI & National Medical Research Centre for Obstetrics, Gynecology and Perinatology Named after Academician V.I.Kulakov of the Ministry of Health of the Russian Federation, Moscow, Russia 

^d Department of Pathology, School of Medicine, Aristotle University of Thessaloniki, Greece

^e Urology Department, Hospital CUF Porto, Porto, Portugal

^f Escola de Medicina e Ciências Biomédicas, Universidade Fernando Pessoa, Porto, Portugal

 

Abstract:

Prostate cancer is a prevalent male malignancy, with increasing incidence rates placing significant diagnostic burdens on pathology services worldwide. Artificial intelligence (AI) is emerging as a promising aid in enhancing diagnostic efficiency and accuracy. This study evaluates the clinical benefits of AI-assisted prostate biopsy (PB) diagnosis, with Paige Prostate tool, compared to non-AI-assisted PB diagnosis, focusing on its predictive accuracy for features in radical prostatectomy (RP) specimens. A retrospective analysis included 55 patients divided into two cohorts: one with non-AI-assisted PB diagnosis (n = 25) and another with AI-assisted PB diagnosis (n = 30). Pathological assessments recorded tumor size, Gleason score, Grade Group, and perineural invasion. The correlation between PB and RP results was analyzed, with statistical significance set at p < 0.05. AI-assisted PB diagnosis showed faster reporting times by 24 hours, enhancing workflow efficiency. AI assistance improved the correlation of tumor size between PB and RP, showing a substantial agreement (R=0.646, p < 0.001) compared to non-AI (R=0.479, p = 0.015). Gleason Score concordance increased by 13 % in the AI-assisted group, achieving 73.3 % versus 60 % in the non-AI-assisted group. This small pilot study suggests that AI-assisted PB diagnosis appears to enhance efficiency and accuracy in the diagnosis of prostate cancer, a finding to be confirmed with further studies.

 

Revista: Pathology-Research and Practice

 

Link: https://lnkd.in/d3Zzc29h