European Radiology: How to achieve undifferentiated MTA assessment among radiologists with different clinical experience?

Medial temporal atrophy (MTA) is an important structural finding associated with Alzheimer's disease, but MTA can also be present in preclinical dementia
.
In clinical practice, the interpretation of structural imaging is usually based on a radiologist's visual assessment
.
It has been shown that structured reporting using the Visual Rating Scale (VRS) can increase the frequency of MTA reporting with little effect on accuracy, while also affecting the diagnostic outcome of cognitive impairment
.

A recent European survey showed that 75% of institutions use VRS in clinical practice, of which 82% frequently use the White Matter Change Scale and 81% use the MTA Scale; other VRS are less common
.
Lack of training is the main reason for not using VRS in clinical practice
.
Some scholars believe that the use of VRS in practice will improve the accuracy of diagnosis .
However, no studies have compared ratings to the gold standard rating of MTA when assessing the accuracy and agreement between models of MTA .

A study published today in the journal European Radiology assessed the inter-modal consistency of MTA scores in a non-demented population and the accuracy between radiologists with different clinical experience , making it possible to standardize the process of MTA identification and assessment.
.
 

Four raters (two junior radiologists and two senior neuroradiologists) scored MTA on CT and MRI scans using Scheltens' MTA scale
.
Scores were compared with the consensus scores of two experienced neuroradiologists to estimate true positive and negative rates (TPR and TNR) and MTA overestimation and underestimation
.
Inter-model agreement expressed as Cohen's κ (dichotomous data), Cohen's κw, and two-way mixed, single-measure, agreement ICC (ordinal data) was determined
.
Adequate agreement was defined as κ/κw ≥ 0.
80 and ICC ≥ 0.
80 (95% CI significance level ≥ 0.
65)
.
 

Forty-nine subjects with cognitive impairment (median age 72 years, 27% abnormal MTA) were included in the study
.
Only primary radiologists achieved sufficient agreement, expressed in Cohen's κ
.
All raters achieved sufficient agreement, expressed as Cohen's κw and ICC
.
True positivity rates ranged from 69% to 100%, and TNRs ranged from 85% to 100%
.
No underestimation or overestimation of MTA was observed
.
There were no differences in scores between radiologists
.
 

FIG prepared according Scheltens image MTA levels of 5 points example ) .
Each side is scored individually and an overall maximum score is given .
MTA 0=no atrophy, MTA 1=mild widening of the choroidal fissure, MTA 2=moderate widening of the choroidal fissure, mild widening of the temporal horn, and mildly decreased hippocampal height .
MTA 3 = severely widened choroidal fissure, moderately widened temporal horn, moderately decreased hippocampal height, MTA 4 = severely widened choroidal fissure, severely widened temporal horn, severely decreased hippocampal height .
 

The present study shows that when the Scheltens MTA scale was used to score MTA in non-demented populations, radiologists with different clinical experience reached sufficient agreement across different modalities with similar accuracy
.
However, TPR differed between raters, possibly due to differences in scoring >
.