Establishing Automatic Method of Counting and Classify Bone Marrow and Peripheral Blood Cells

Brief Summary: Counting and classification of blood cells in a bone marrow smear and peripheral blood smear are essential to clinical hematology. To this date, this procedure has been carried out in a manual manner in the great majority of clinical settings. There is often inconsistency in the counting result between different operators largely due to its manual nature. There has not been an effective and standard method for blood smear preparation and automatic counting and classification. The recent advent of deep neural network for medical image processing introduced new opportunities for an effective solution of this long-standing problem. Numerous results have been published on the effectiveness of convolutional neural network in clinical image recognition task.

Recruitment & Eligibility

Inclusion Criteria

Patients who have suspected or confirmed hematological diseases and receive bone
marrow or peripheral blood cell morphological examination in National Taiwan University Cancer Center
Patients who are aged more than 20 y/o

Exclusion Criteria

•Patients who are not willing to sign informed consents

Primary Outcome Measures

Name	Time	Method
Evaluate the accuracy of cell counting and classifying between automatic method and manual method through digital microscopic photos of bone marrow smear and peripheral blood smear using deep convolutional neural networks	3 years

Secondary Outcome Measures

Name	Time	Method

Locations (1): National Taiwan University Tai-Chen Cell Therapy Center
🇨🇳
Taipei, Taiwan

Receive instant email notifications about changes in this clinical trial

Receive instant email notifications about changes in this clinical trial