European Radiology: Deep Learning + Radiomics!

 Soft tissue sarcoma (STS) originates from mesenchymal tissue, with complex histological components and diverse clinical manifestations, accounting for 1% of the total adult cancer incidence
.

At this stage , radical resection is the preferred treatment for STS patients .

 Soft tissue sarcoma (STS) originates from mesenchymal tissue, with complex histological components and diverse clinical manifestations, accounting for 1% of the total adult cancer incidence

Patients with STSs are often treated with neoadjuvant radiotherapy or chemotherapy after surgical resection to improve the local control rate and prolong their survival time

The American College of Radiology's appropriateness criteria recommend MRI for monitoring local recurrence of malignant or invasive soft tissue tumors

Recently, a study published in the journal European Radiology established and tested a model using DL and HCR radiomics signatures to predict the risk of recurrence in patients with STS undergoing surgical resection, while incorporating routinely applied staging systems and other studies.
Statistical models were incorporated into model comparisons to evaluate the efficacy of this study model in recurrence prediction, and to provide technical support for clinically formulating personalized treatment plans and improving patient prognosis .

In total , this study evaluated 282 patients undergoing MRI scans and STS surgical resection at three independent centers .
In addition, 113 of the 282 patients underwent additional contrast- enhancing MRI scans .
We divided participants into a developmental cohort and an external testing cohort .
The developmental cohort consisted of patients from one center, and the external testing cohort consisted of patients from two other centers .
We established two MRI-based DLRNs for predicting tumor recurrence after STS resection .
We tested DLRNs universally and compared them with other predictive models constructed by using widely adopted predictors (i.
e.

The DLRN1 model incorporates plain MRI-based radiomic features into clinical data, and the DLRN2 model combines radiomic features extracted from plain and enhanced MRI with clinical predictors

 

Fig.
Prediction of cumulative recurrence rate bydeep learning radiology nomograms (DLRNs) for three different risk stratifications
.

Fig.
Prediction of cumulative recurrence rate bydeep learning radiology nomograms (DLRNs) for three different risk stratifications
.

Fig.

This study constructed two radiomics nomograms based on clinical and radiomic features .

Original source:

Shunli Liu,Weikai Sun,Shifeng Yang,et al.

Deep learning radiomic nomogram to predict recurrence in soft tissue sarcoma: a multi-institutional study.

Shunli Liu,Weikai Sun,Shifeng Yang,et al.
Deep learning radiomic nomogram to predict recurrence in soft tissue sarcoma: a multi-institutional study.
DOI:10.
1007/s00330-021-08221-0Leave a comment here