Diagnostic Efficiency of Artificial Intelligence for Surgical Neuropathology

Jinsong Wu

Status: Unknown

Conditions: Central Nervous System Neoplasms

Treatments: Diagnostic Test: Artificial Intelligence; Diagnostic Test: Practicing Pathologists; Diagnostic Test: Gold Standard

Study type: Interventional

Funder types: Other

Identifiers

NCT04671368

PAAI2020

Details and patient eligibility

About

This is a multi-center, prospective, self-controlled, diagnostic accuracy comparative study of Artificial Intelligence Diagnostic System for Surgical Neuropathology. The investigators will compare the diagnostic efficiency of Artificial Intelligence with that of practicing pathologists, and suppose that the diagnostic efficiency of artificial intelligence in prospective clinical data is no less than that of pathologists.

Full description

In this study, 141 patients will be recruited. After being enrolled, the patients will accept surgery and specimens for pathological analysis will be taken according to the routine treatment process.

The histopathologic slides will then be digitized by a whole-slide scanner. The images will be reviewed by gold standard committee for evaluation of ground truth. And then be separately diagnosed by Artificial Intelligence Diagnostic System and practicing pathologists. So the investigators can compare the diagnostic efficiency of Artificial Intelligence with that of pathologists, thus understand the gap between artificial intelligence and actual clinical practice.

Enrollment: 141 estimated patients

Sex: All

Ages: 18+ years old

Volunteers: No Healthy Volunteers

Inclusion criteria

1. Patients or their guardians understand the research process, agree to use their data, and sign the informed consent form;
2. Aged >=18 years;
3. MRI shows intracranial spaceoccupying lesions;
4. The clinical diagnosis is glioma, metastasis or lymphoma thus requiring surgical treatment;
5. The patient is willing to accept the surgery.

Exclusion criteria

1. The patient has serious underlying diseases thus is not suitable for surgery;
2. After further clinical evaluation, surgical treatment was not the best choice;
3. The patient participate in clinical research of other drugs or devices;
4. The researchers believe that there are other factors that will make the patients unable to complete the study.

Trial design

Primary purpose: Diagnostic

Allocation: Non-Randomized

Interventional model: Parallel Assignment

Masking: Single Blind

---

141 participants in 3 patient groups

Artificial Intelligence

Experimental group

Description:

A deep learning based artificial intelligence diagnostic system(DOI:10.1093/neuonc/noaa163)

Treatment:

Diagnostic Test: Artificial Intelligence

Practicing Pathologists

Active Comparator group

Description:

One pathologist who has at least 5 years of experience

Treatment:

Diagnostic Test: Practicing Pathologists

Gold Standard

Other group

Description:

A committee composed of two expert pathologists who has at least 10 years of experience and one expert pathologist who has at least 15 years of experience

Treatment:

Diagnostic Test: Gold Standard