Sequential Analysis of Murine Myelofibrosis Models Using a Novel Deep Learning-Based Fibrosis Quantitative Method.

Area of Science:

• Biomedical Engineering
• Computational Biology
• Oncology

Background:

• Primary myelofibrosis (MF) diagnosis relies on qualitative assessments, hindering early detection of disease progression.
• Current methods struggle to identify subtle changes in reticular fibers (RFs) during early stages of MF.
• There is a need for quantitative methods to evaluate MF and its progression.

Purpose of the Study:

• To develop and validate a novel deep learning-based method for quantitative assessment of RFs in MF.
• To evaluate the potential of this method in tracking temporal changes in fibrosis, splenomegaly, and hematopoietic stem cell dynamics.
• To correlate quantitative fibrosis measurements with clinical data, including MF grade and genetic mutations.

Main Methods:

• A deep learning model was developed to quantitatively measure RFs as an indicator of MF.
• The method was applied to analyze fibrosis in bone marrow and spleen in two MF mouse models (romiplostim-induced and Jak2V617F-transformed).
• Quantitative fibrosis measurements were correlated with clinical data from patients with myeloproliferative neoplasms.

Main Results:

• The deep learning method successfully captured temporal changes in bone marrow and spleen fibrosis.
• In Jak2V617F mice, splenomegaly and extramedullary hematopoiesis preceded MF.
• Quantitative fibrosis measurements showed a significant correlation with MF grade and JAK2V617F mutant allele burden.

Conclusions:

• The developed deep learning method offers a quantitative approach for assessing MF.
• This novel method has the potential for clinical application in early detection and monitoring of MF progression.
• Quantitative fibrosis assessment may improve diagnostic accuracy and treatment evaluation in MF.