Monitoring computed tomography (CT) scan image

Inventors

Warier, Prashant • Modi, Ankit • Putha, Preetham • Vanapalli, Prakash • Challa, Vikash

Assignees

Qure AI Technologies Pvt Ltd

Abstract

Disclosed is a system and a method for monitoring a CT scan image. A CT scan image may be resampled into a plurality of slices using a bilinear interpolation. A region of interest may be identified on each slice using an image processing technique. The region of interest may be masked on each slice using deep learning. Subsequently, a nodule may be detected as the region of interest using the deep learning. Further, a plurality of characteristics associated with the nodule may be identified. Furthermore, an emphysema may be detected in the region of interest on each slice. A malignancy risk score for the patient may be computed. A progress of the nodule may be monitored across subsequent CT scan images. Finally, a report of the patient may be generated.

Core Innovation

The invention provides a method for monitoring a Computed Tomography (CT) scan image of a patient. The method receives a CT scan image, applies gaussian smoothing to counteract noise, and resamples the CT scan image into a plurality of slices using bilinear interpolation. The method identifies a region of interest on each slice using an image processing technique and masks the region of interest by removing black or air areas and fatty tissues around the region of interest using deep learning.

Using the masked region of interest, the method detects a nodule as the region of interest with deep learning. The method determines a plurality of characteristics associated with the nodule using the image processing technique, where the plurality of characteristics comprise a diameter, a calcification, a lobulation, a spiculation, a volume, and a texture. The method also detects an emphysema in the region of interest on each slice using the deep learning.

The method computes a malignancy risk score for the patient in real-time based on the plurality of characteristics and a trained data model. The trained data model comprises historical data related to different diameter of nodules, different calcification of nodules, different lobulation of nodules, different spiculation of nodules, different volume of nodules, and different texture of nodules, and the malignancy risk score is dependent on a weightage of each characteristic. The method monitors a progress of the nodule in real-time over a predefined time period across subsequent CT scan images based on the diameter, a total volume of the nodule and the malignancy risk score, and generates a report that includes the nodule, the emphysema, the malignancy risk score, the progress of the nodule and a follow-up check with a health practitioner, thereby monitoring the CT scan image.

Claims Coverage

The independent claims cover three related aspects: a monitoring method for CT scan images, a monitoring system implementing the same workflow, and a non-transitory computer program product that stores instructions to perform the same workflow. Across these independent claims, seven inventive features are centered on deep learning-based region-of-interest masking, nodule and emphysema detection, characteristic extraction, a real-time malignancy risk score based on a trained data model with weighted characteristics, and real-time progress monitoring over a predefined time period with report generation including follow-up.

The independent claims collectively cover an end-to-end CT monitoring workflow that preprocesses a CT scan image, uses deep learning to mask a region of interest and detect a nodule and emphysema, derives multiple nodule characteristics, computes a real-time malignancy risk score from a trained data model with weighted characteristics, monitors nodule progress across subsequent CT scan images over a predefined time period, and generates a patient report including follow-up with a health practitioner.

Stated Advantages

Documented Applications

No documented applications found