Cerebral staging of lung cancer: is one single contrast-enhanced T1-weighted three-dimensional gradient-echo sequence sufficient?

Diagnostic performance of axial T2-weighted MRI sequence for exclusion of brain tumour in paediatric patients with non-localizing symptoms

OBJECTIVE

To establish diagnostic performance of a single axial T2-weighted sequence for detection of brain tumours in children with non-localizing symptoms, compared to a standard MRI protocol.

METHODS

Retrospective analysis of children undergoing MRI brain imaging for suspected brain tumours with non-localizing symptoms over a 3-year period. Axial T2-weighted images were blindly reviewed by 2 experienced paediatric neuroradiologists. Primary analysis was calculation of diagnostic performance metrics for tumour identification using axial T2-weighted image only compared to the standard MRI protocol.

RESULTS

For 312 children undergoing MRI brain during the study period, sensitivity and specificity for brain tumour detection based on axial T2-weighted images in children with non-localizing symptoms were 1.000 (95% CIs 0.598, 1.000) and 0.998 (95% CI 0.990, 0.999), respectively. Based on T2-weighted images alone, 50 patients (16%) were flagged as needing recall for further imaging compared to 14 (4.5%) recalled after the standard protocol.

CONCLUSIONS

Axial T2-weighted images have high sensitivity and specificity for detection of brain tumours in children with non-localizing symptoms but are associated with increased imaging recall rates. Prospective evaluation of this approach to identify patients requiring more comprehensive imaging is warranted.

ADVANCES IN KNOWLEDGE

A truncated MRI protocol with single axial T2-weighted sequence has high diagnostic performance for brain tumour detection in children with non-localizing features. Radiologists can be reassured that a child with this presentation who is unable to complete the full MRI scan protocol is very unlikely to have a brain tumour missed provided an axial T2-weighted sequence is obtained.

Location of core diagnostic information across various sequences in brain MRI and implications for efficiency of MRI scanner utilization

OBJECTIVE

Targeting redundancy within MRI can improve its cost-effective utilization. We sought to quantify potential redundancy in our brain MRI protocols.

MATERIALS AND METHODS

In this retrospective review, we aggregated 207 consecutive adults who underwent brain MRI and reviewed their medical records to document clinical indication, core diagnostic information provided by MRI, and its clinical impact. Contributory imaging abnormalities constituted positive core diagnostic information whereas absence of imaging abnormalities constituted negative core diagnostic information. The senior author selected core sequences deemed sufficient for extraction of core diagnostic information. For validating core sequences selection, four readers assessed the relative ease of extracting core diagnostic information from the core sequences. Potential redundancy was calculated by comparing the average number of core sequences to the average number of sequences obtained.

RESULTS

Scanning had been performed using 9.4±2.8 sequences over 37.3±12.3 minutes. Core diagnostic information was deemed extractable from 2.1±1.1 core sequences, with an assumed scanning time of 8.6±4.8 minutes, reflecting a potential redundancy of 74.5%±19.1%. Potential redundancy was least in scans obtained for treatment planning (14.9%±25.7%) and highest in scans obtained for follow-up of benign diseases (81.4%±12.6%). In 97.4% of cases, all four readers considered core diagnostic information to be either easily extractable from core sequences or the ease to be equivalent to that from the entire study. With only one MRI lacking clinical impact (0.48%), overutilization did not seem to contribute to potential redundancy.

CONCLUSION

High potential redundancy that can be targeted for more efficient scanner utilization exists in brain MRI protocols.