Development of Artificial Intelligence Models for Segmentation and Characterization of Prostate Cancer: a Single-center Retrospective Observational Study.

NCT06168864 | Completed

• 1 other identifier: AI_Pca
• Study Type: observational
• Target: 350
• Locations: 1 country
• Sites: 1

Brief Summary

Prostate cancer is the second most common cancer in the male population. This pathology represents an oncological and public health problem especially in developed countries, due to a greater presence of elderly men in the population. Medical imaging plays a central role in the staging and restaging of prostate disease. Magnetic resonance imaging (MRI), computed tomography (CT) and positron emission tomography (PET) are among the methods commonly used in normal clinical practice for the characterization of prostate cancer. To date, the study of these images is limited to a qualitative visual analysis, however there is increasing evidence relating to the usefulness of introducing a quantitative (or semi-quantitative) analysis of biomedical images. The current increase in available imaging data, and their quality, allows the application of artificial intelligence methods also in the medical field for the automation of tasks (e.g. automatic segmentation) and classification (e.g. tumor aggressiveness). The extraction of quantitative data, and more generally the study of tumor lesions, requires manual segmentation by one or more doctors. This process requires very long times as each image must be processed individually; furthermore, the result also depends on the level of experience of the doctor carrying out the segmentation and this could create a source of heterogeneity, affecting the reproducibility of the segmentation. AI-based automatic segmentation methods can be applied to medical images for the localization of tumor lesions, thus exceeding the limits of manual segmentation.

Conditions

Prostate Cancer

Outcome Measures

Primary Outcomes (1)

• Artificial intelligence algorithms for the classification of prostate cancer lesions on medical images.

  PET images from enrolled patients will be used to create models that investigate the ability of artificial intelligence to automate tumor segmentation tasks.

  2 years

Interventions

Artificial intelligence models for segmentation and characterization of prostate cancer DIAGNOSTIC_TEST

rtificial intelligence algorithms for the automatic segmentation of prostate cancer lesions on medical images.

Eligibility Criteria

• Age: 18 Years+
• Sex: male
• Age Groups: Adult (18-64), Older Adult (65+)
• Sampling Method: Probability Sample

Study Population

Patients with prostate cancer undergoing PET examination with 68 Ga-PMSA (PET/CT or PET/MRI), since 01/06/2020, at the U.O. of Nuclear Medicine at the San Raffaele Hospital on the clinical indication of the specialist.

You may qualify if:

• Patients with histological diagnosis of prostate cancer;
• Patients who performed a PET exam with 68 Ga-PMSA.

You may not qualify if:

• CT and MR images with artifacts that preclude interpretation of results.

Sponsors & Collaborators

• IRCCS San Raffaele: lead

Study Sites (1)

Irccs San Raffaele

Milan, 20132, Italy

Location

MeSH Terms

Conditions

Prostatic Neoplasms

Condition Hierarchy (Ancestors)

Genital Neoplasms, Male; Urogenital Neoplasms; Neoplasms by Site; Neoplasms; Genital Diseases, Male; Genital Diseases; Urogenital Diseases; Prostatic Diseases; Male Urogenital Diseases

Study Design

• Study Type: observational
• Observational Model: COHORT
• Time Perspective: RETROSPECTIVE
• Sponsor Type: OTHER
• Responsible Party: PRINCIPAL INVESTIGATOR
• PI Title: Professor in Diagnostic Imaging and Radiotherapy Faculty of Medicine and Surgery, Vita-Salute San Raffaele University Director, Department of Nuclear Medicine, IRCCS Ospedale San Raffaele

Study Record Dates

• First Submitted: December 5, 2023
• First Posted: December 13, 2023
• Study Start: January 6, 2020
• Primary Completion: June 1, 2022
• Study Completion: June 1, 2022
• Last Updated: December 13, 2023

Record last verified: 2023-12

Data Sharing

• IPD Sharing: Will not share

Locations

IT (1)