European Radiology: Radiographic, noninvasive grading diagnosis of mandibular squamous cell carcinoma

At present, multi-probe computed tomography (MDCT) and magnetic resonance imaging (MRI) are the most important imaging methods for preoperative evaluation of
oral cancer.
Although magnetic resonance imaging provides superior soft-tissue contrast, longer scan times are a major limitation
associated with this technique.
In addition, although volume assessment greatly increases diagnostic accuracy, two-dimensional (2D) scanning of the oral and maxillofacial area in clinical practice is often affected
by metal artifacts caused by sports and dental restorations.
This is because three-dimensional (3D) scanning requires longer scan times and the gradient echo (GRE) sequences commonly used for 3D scanning are susceptible to magnetic field inhomogeneities
.

Compressed sensing (CS) is a recent technological breakthrough that accelerates MRI acquisition, allowing data to be reconstructed from highly undersampled measurements that can be efficiently utilized under conditions of sparse, pseudo-random undersampling and nonlinear reconstruction
.

It has recently been reported that the three-dimensional T1 turbine field echo combined with compressed sensing (CS-3D-T1TFE) sequences is of great
significance for oral and maxillofacial MRI scanning.
This technique reduces acquisition time to less than 30% of the time required for two-dimensional multilayer spin-echo (2D-MS-SE) sequences without degradation of image quality, maintaining equal or higher signal-to-noise ratio (SNR) and image quality
.
Recently, a study published in the journal European Radiology compared CS-3D-T1TFE images and MDCT images for the classification of jaw cancer, and compared these two sets of images with histopathological results as the gold standard to verify the value of
CS-3D-T1TFE reconstruction images.

This study retrospectively included 24 patients with mandibular squamous cell carcinoma (SCC), each of whom underwent MRI prior to surgery
, including CS-3D-T1TFE and MDCT examinations.
For both examinations, a 0.
5 mm thick coronal surface and a 0.
5 mm layer thick planar image perpendicular to and parallel to the tooth was constructed.

Two radiologists scored bone invasion on cortical bone, cancellous bone, and mandibular canal (MC) and assessed the agreement
among raters by weighted kappa statistics.
Of the 24 patients who underwent surgery, the correlation between bone violations and histopathological evaluation by pathologists was assessed using the Pearson correlation coefficient
.
Soft-tissue invasion is assessed by diagnosing invasion of the hyoid muscles, gingival folds, and chewing space, and assessing inter-observer agreement
using kappa statistics.

Observer-to-observer agreement for bone invasion assessment of CS-3D-T1TFE was almost perfect
.
In 20 cases, the image assessment of two observers was consistent with the pathological assessment, showing a high correlation (r>0.
8).

CS-3D-T1TFE also showed higher rate-to-rate agreement
with all measures of soft tissue invasion than MDCT.

An 85-year-old woman with squamous cell carcinoma of the left mandible (T2N0M0) (case 9).

a Coronal section is reconstructed from CS-3D-T1TFE images, which were acquired
within 1 min and 43 sec after angiography.
b The 2D-MS-SE coronary chart was acquired at approximately 7 minutes and 30 seconds within 6 minutes after imaging; The layer thickness is 3.
0mm
.
In A, the tumor is far away from the mandible canal (MC), while in B, the tumor appears to infiltrate the MC

This study showed that the reconstructed CS-3D-T1TFE images had high
diagnostic value for jaw cancer.
For the evaluation of bone invasion as well as soft tissue invasion, the agreement between scorers of imaging assessments of CS-3D-T1TFE images was higher than that of MDCT and showed a high correlation
with pathological evaluation results.

Original source:

Yukiko Kami,Toru Chikui,Osamu Togao,et al.
Usefulness of reconstructed images of Gd-enhanced 3D gradient echo sequences with compressed sensing for mandibular cancer diagnosis: comparison with CT images and histopathological findings.
DOI:10.
1007/s00330-022-09075-w