Aims and objectives
Ki-67 index is a proliferation index marker,
which provides crucial information about the mitotic activity,
and the growth rate of the tumor.Thus,
it plays an important role as a prognostic factor in breast cancereffecting the management decisions.
As aproliferation index marker,
Ki-67 plays an important role in categorizing the disease within themolecular subtypes and guides therapy related decision making.
Non-invasive prediction models for the prognostic factors including Ki-67 have been largely based on breast MRI findings.
As it is well known from the previous studies,...
Methods and materials
In this IRB approved study,
Out of 126 consecutive patients' retrospective data with written consent,
89 patients (mean±std age 49.82±10.91 and age range [28-82] yrs.) with full histopathological and MRI data were selected.
In surgical histopathological data ER(+) PR(+) HER(-) was classifiedas Luminal A,
ER(+) PR(-) HER(±) was classified as luminal B,
ER(-) PR(-) HER2(+) was classified as HER2-enriched and ER(-) PR(-) HER2(-) was classified as triple negative.
The cohort revealed 65 luminal A,
18 luminal B,
4 triple negative cases and 2 HER2-enriched occurrences...
Results
With the exception of some of the variants of K-nearest neighbor (KNN) and Boosted Trees algorithms,
all of the remaining algorithms provide accuracies between 73%-86.5%.
This is mainly due to skewed distribution of the cohort i.e.
65/89 Lum A vs.
24/89 non-Lum A patients which pulls the discriminant based algorithms towards misclassifications.
However,
Fine KNN,
Weighted KNN and Boosted Trees prove to be very efficient with the accuracies of 98.9%.
While Fine KNN misclassified only one Lum A patient,
the remaining two algorithms misclassified one...
Conclusion
The contrast between the outcomes of set based (e.g.
KNN) and separator based (e.g.
support vector machines) algorithms demonstrate that ADC-Ki67 space might potentially provide relevant information with the use of nonlinear and set based machine learning algorithms.
References
Pinker K,
Chin J,
Melsaether AN,
Morris EA,
Moy L.
Precision Medicine and Radiogenomics in Breast Cancer: New Approaches toward Diagnosis and Treatment.
Radiology 2018; 287(3): 732-747.
Basser PJ.
Inferringmicrostructuralfeaturesand thephysiologicalstateoftissuesfromdiffusion-weighted images.
NMR Biomed.1995 Nov-Dec;8(7-8):333-44.
Leithner D,
Wengert GJ,
Helbich TH,
Thakur S,
Ochoa-Albiztegui RE,
Morris EA,
Pinker K.
Clinical role of breast MRI now and going forward.
Clin Radiol.
2017 Dec 8.
pii: S0009-9260(17)30521-4.
Yoo EY,
Nam SY,
Choi HY,
Hong MJ.
Agreement between MRI and pathologic analyses for determination of tumor size and...