Machine Learning Rules Out Lymph Node Metastasis

machine learning
Source: Medical Xpress

Approximately 70% of patients are aware of when their cancer spreads to the lymph nodes without an invasive sentinel node biopsy. New research published in the Journal of Nuclear Medicine shows that with machine learning (AI type), metastasis of the axillary lymph node can be ruled out using imaging with PET/MRI.

Moreover, lymph node metastasis is crucial when treatment planning is considered. Especially based on the extent of the surgery and radiation. Hence it is of high relevance clinically for distinguishing patients without lymph node metastasis.

Author of the study, Janna Morawitz MD says,

Sixty percent of patients do not have lymph node metastases at initial diagnosis of breast cancer

As such, it would be desirable to be able to prove negative lymph node status by imaging with a high degree of certainty to spare these patients the invasive procedure of biopsy or surgery.

Researchers aim to determine if machine learning prediction models can determine lymph node status in PET/MRI examination accurately just like radiologists. A total of 303 patients with breast cancer were recruited for the study. Furthermore, they were divided into two groups: a training sample and a testing sample.

All the patients went through MRI and whole-body 18F-FDG PET/MRI. Additionally, the imaging sets of data were evaluated for metastasis of the lymph node. It was based on functional and structural features. However, the models were made using the MRI and PET/MRI training group samples, which were then applied to the testing group samples.


The accuracy of the diagnosis using MRI was the same 87.5% for both radiologists and machine learning algorithms. Moreover, for PET/MRI it was 89.3% for radiologists and 91.2% for machine learning. Other than this, when the machine learning model was adjusted for PET/MRI, sensitivity was 96.2% and specificity was 68.2%.

Morawitz noted,

Based on the information gleaned from the MRI and PET/MRI scans, decision trees can be made to help radiologists—especially younger radiologists—determine whether a sentinel lymph node biopsy is warranted,

Incorporating this model into daily practice could potentially replace sentinel lymph node biopsy in the future.


Please enter your comment!
Please enter your name here