Systems and methods for detection of infectious respiratory diseases
Inventors
Putha, Preetham • Tadepalli, Manoj • Reddy, Bhargava • Raj, Tarun • Jagirdar, Ammar • Rao, Pooja • Warier, Prashant
Assignees
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-11308612-B2
Publication Date
2022-04-19
Expiration Date
Abstract
This disclosure generally pertains to systems and methods for detection of infectious respiratory diseases by implementation of an automated X-rays-based triage approach alongside algorithmic clinical sample pooling for molecular diagnosis. Certain embodiments relate to methods for the development of deep learning algorithms that perform machine recognition of specific features and conditions in chest X-ray imaging data. The chest X-ray imaging data is used to guide the pooling strategy of clinical samples for a molecular test.
Core Innovation
The invention relates to detection of an infectious respiratory disease by implementing an automated X-rays-based triage approach together with algorithmic clinical sample pooling for molecular diagnosis. A method receives a clinical sample and chest X-ray scan imaging procedure data from a subject, processes images from the chest X-ray scan imaging procedure data, and uses a deep learning system to detect and localize medical abnormalities. The deep learning system outputs a bounding box representing the precise location and extent of the medical abnormalities and provides recognition scores for each abnormality.
From the medical abnormality recognition scores, the method generates a second score for each subject that corresponds to a level of recognition of an infectious respiratory disease based on abnormalities that suggest typical or atypical symptoms of the infectious respiratory disease. The second score is mapped to a probability of a subject having the infectious respiratory disease. Based on this probability, the clinical sample is ranked from lowest to highest, and a pooling size is selected for a molecular test using a collective probability of all subjects in a pool.
The collective probability is defined as a sum of the probability of each subject having the infectious respiratory disease in the pool, and the collective probability is constrained to be lesser than or equal to 1. A corresponding system implementation includes a first subsystem that automates detection and localization of medical abnormalities on chest X-ray imaging scans using a deep learning algorithm, and a second subsystem that algorithmically pools clinical samples from each subject based on the mapped subject-level probability and the defined collective probability constraint.
Claims Coverage
The document includes two independent claims. They share a common set of inventive features: deep-learning detection and localization of radiographic medical abnormalities on chest X-ray scans with bounding boxes and recognition scores, conversion of abnormality scores into a subject-level infectious respiratory disease score and probability, and subject ranking and selection of a pooling size for a molecular test using a collective pool probability defined as a sum of subject probabilities constrained to be ≤ 1.
Automated X-rays-based triage with deep learning detection and localization
A method/system receives chest X-ray scan imaging procedure data from a subject, processes images, and detects and localizes medical abnormalities using a deep learning system that outputs a bounding box representing the precise location and extent of the medical abnormalities together with a first score corresponding to a level of recognition of each medical abnormality.
Infectious respiratory disease scoring mapped to subject probability
The method/system generates a second score for each subject corresponding to a level of recognition of an infectious respiratory disease based on the first scores of medical abnormalities that suggest typical or atypical symptoms of the infectious respiratory disease, and maps the second score to a probability of the subject having the infectious respiratory disease.
Ranking subjects by probability and selecting pooling size for molecular diagnosis with collective probability constraint
The method ranks clinical samples from lowest to highest based on the second score, and selects a pooling size for a molecular test based on a collective probability of all subjects having the infectious respiratory disease in a pool, where the collective probability is a sum of the probability of each subject in the pool and where the probability is lesser than or equal to 1.
Overall, the claim set covers combining automated X-rays-based triage that converts abnormality recognition into subject-level infectious respiratory disease probabilities with algorithmic clinical sample pooling for molecular diagnosis, where pooling selection is governed by a collective pool probability defined as a sum of subject probabilities and constrained to be ≤ 1. Dependent claim refinements specify additional architectural and disease/assay details, including pooling via convex approximation of LogSumExp, an SE-ResNeXT-50 abnormality detector, and molecular diagnosis via real-time PCR for semiquantitative detection including Mycobacterium tuberculosis and rifampin resistance.
Stated Advantages
Not explicitly described in patent.
Documented Applications
Not explicitly described in patent.
Interested in licensing this patent?