A Diagnostic Algorithm using Multi-parametric MRI to Differentiate Benign from Malignant Myometrial Tumors: Machine-Learning Method

Abstract

This study aimed to develop a diagnostic algorithm for preoperative differentiating uterine sarcoma from leiomyoma through a supervised machine-learning method using multi-parametric MRI. A total of 65 participants with 105 myometrial tumors were included: 84 benign and 21 malignant lesions (belonged to 51 and 14 patients, respectively; based on their postoperative tissue diagnosis). Multi-parametric MRI including T1-, T2-, and diffusion-weighted (DW) sequences with ADC-map, contrast-enhanced images, as well as MR spectroscopy (MRS), was performed for each lesion. Thirteen singular MRI features were extracted from the mentioned sequences. Various combination sets of selective features were fed into a machine classifier (coarse decision-tree) to predict malignant or benign tumors. The accuracy metrics of either singular or combinational models were assessed. Eventually, two diagnostic algorithms, a simple decision-tree and a complex one were proposed using the most accurate models. Our final simple decision-tree obtained accuracy = 96.2%, sensitivity = 100% and specificity = 95%; while the complex tree yielded accuracy, sensitivity and specificity of 100%. To summarise, the complex diagnostic algorithm, compared to the simple one, can differentiate tumors with equal sensitivity, but a higher specificity and accuracy. However, it needs some further time-consuming modalities and difficult imaging calculations. Trading-off costs and benefits in appropriate situations must be determinative.

Radiomics and Artificial Intelligence in Uterine Sarcomas: A Systematic Review

“…Malek [16] A simple algorithm showed 96.2% accuracy, 100% sensitivity and 95% specificity. The complex algorithm yielded accuracy, sensitivity and specificity of 100%.…”

“…Ten studies were duplicates across PubMed and Scopus and thus eliminated, resulting in 744 studies to be screened. According to the previously described inclusion and exclusion criteria, 722 papers were excluded, and 6 full-text articles were included in this systematic review [16][17][18][19][20][21]. Details about the literature search results are reported in Figure 1.…”

“…Although a complex algorithm showed 100% sensitivity, specificity and accuracy in the differentiation of myomas from leiomyosarcomas, this classifier was deemed too complicated for routine clinical practice [16,20]. The models based on radiomics features extracted from the whole uterus outperformed the ones based on features extracted only from the macroscopic tumour or from the tumour and a small region of the surrounding tissue [18].…”

Advances in the Preoperative Identification of Uterine Sarcoma

“…TCR combined with apparent diffusion coefficient (ADC) can completely differentiate sarcomas from uterine fibroids. Several studies [ 54 , 55 , 56 , 57 ] have found that diffusion-weighted imaging (DWI) high signal and low ADC values were predictors of uterine sarcoma, whereas slightly different ADC values were determined by other studies. DWI and ADC combined with lymph node enlargement or retroperitoneal masses [ 58 ] can improve the detection of uterine sarcomas.…”

“…For uterine tumors with high T2 signal, radiomics combined with clinical variable models [ 76 , 77 ] can achieve predictive performance higher than that of imaging physicians. Malek et al [ 57 ] obtained the best simple decision tree with 96.2% accuracy, 100% sensitivity, and 95% specificity by analyzing 13 features of multiparametric MRI with a machine-learning algorithm, whereas the accuacy, sensitivity, and specificity of the complex tree were up to 100%. Good progress has also been made in radiomics based on ADC maps to differentiate uterine sarcomas from uterine leiomyomas [ 78 , 79 ].…”