Systems and related methods of imaging and classifying cells in a blood sample
Inventors
CORREIA DE MATOS NOLASCO LAMAS, Francisco • MANUEL DE OLIVEIRA GARCIA DA FONSECA, João • DO ROSÁRIO MAGALHÃES, André • Pedro Santos Manso Co̊rte-Real, José • Cabeça, Ricardo Manuel Marques Caleiro
Assignees
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-10684206-B2
Publication Date
2020-06-16
Expiration Date
Abstract
Systems and methods for classifying blood cells in a blood sample are disclosed. A series of frames of the blood sample as it flows through a field of view of an image capture device are captured and analysed. Advantageously, the disclosed systems and methods combine the availability of morphological cell data with the convenience of a flow-through arrangement. The classification results can be used for estimating cell counts in a blood sample.
Core Innovation
The invention relates to classifying blood cells in a blood sample by capturing a sequence of frames of the sample as the sample flows through a field of view of an image capture device. Each frame is segmented into objects and background, and each segmented object is tracked over at least some of the frames. The tracked objects are classified in the respective at least some of the frames to obtain a classification result for each tracked object in each of those frames.
The overall classification result for each tracked object is computed based on the respective classification results from the at least some of the frames. The overall classification result classifies each tracked object as one of a plurality of object types, where the object types include one or more cell types. The per-frame classification results are combined so that each tracked object is assigned to an object type that includes one or more cell types.
The overall classification result can be computed using averaging, calculating the median, or applying one or more rules to a sequence of classification results from the at least some of the frames. The classification results for each tracked object can indicate a likelihood that the object belongs to each of the object types. The document also describes estimating a cell count in a blood sample using the same frame-based processing, and determining proportions of objects classified as each cell type relative to the total number of objects classified as one or more cell types.
The document enumerates cell types including neutrophils, basophils, eosinophils, lymphocytes, and monocytes. It further describes a non-isotropic characteristic feature that can be used by thresholding classification results or by thresholding change in the classification results within a preset number of frames.
Claims Coverage
The partial content includes two independent claims: a method for classifying blood cells and a system for estimating a cell count. Across the independent-claim scope, the claims share inventive features centered on frame-based segmentation, object tracking across frames, per-frame object classification, and computing an overall per-tracked-object classification using results from multiple frames; together these features cover classifying each tracked object into a plurality of object types including one or more cell types.
Frame sequence acquisition as sample flows through field of view
Capturing a sequence of frames of the sample with an image capture device as the sample flows through a field of view of the image capture device.
Per-frame segmentation into objects and background
Segmenting each frame into objects and background.
Object tracking across at least some frames
Tracking each segmented object over at least some of the frames.
Per-frame classification of tracked objects
Classifying the tracked objects in each of the respective at least some of the frames to obtain a classification result for each tracked object and each of the respective at least some of the frames.
Overall per-tracked-object classification from multiple frame results
Computing an overall classification result for each tracked object based on the respective classification results from each of the at least some of the frames to classify each tracked object as one of a plurality of object types, wherein the object types include one or more cell types.
System architecture for cell count estimation using the same frame-based processing
An image capture device having a field of view and an image processor configured to capture a sequence of frames, segment each frame into objects and background, track each segmented object over at least some of the frames, classify tracked objects in each of the respective at least some of the frames to obtain classification results, and compute an overall classification result for each tracked object based on the respective classification results to classify each tracked object as one of a plurality of object types including one or more cell types.
Both independent claims cover multi-frame object-based blood-cell classification: segmenting frames, tracking objects across frames, classifying objects per frame, and computing an overall per-tracked-object classification into a plurality of object types including one or more cell types. The system claim applies the same processing to estimate a cell count.
Stated Advantages
Not explicitly described in patent.
Documented Applications
Not explicitly described in patent.
Interested in licensing this patent?