Pediatric CLD encompasses a diverse range of clinicopathological conditions. It has a long-term, slow-progressing course. More than half of all instances of pediatric CLD occur in children under the age of five [1].
HRCT is the preferred imaging modality because of its speed, availability, and familiarity, as well as its ease of use, excellent natural contrast, and spatial resolution of the lung parenchyma. MRI has had limited clinical use due to low signal intensity from the lung parenchyma, as well as the prolonged acquisition time and susceptibility to respiratory motion. The lack of radiation makes pulmonary MRI an excellent primary imaging modality for pediatric evaluations, particularly for patients who require serial and longitudinal follow-up, such as CLD patients. Thankfully, new MRI techniques, such as ultra-short echo time and zero echo time, are increasing the number of clinical applications for pulmonary MRI. These sequences provide a greater SNR from the lung's relatively short T2* due to the use of multi-coil parallel acquisitions and acceleration methods, making pulmonary MRI useful for evaluating lung parenchymal disorders [6].
Different MRI sequence protocols have been raised in the literature. Puderbach et al. provide a comprehensive description of the sequence protocols employed [7]. T2-weighted sequences are the foundation of lung MRI; they are sufficient for morphological diagnosis and are robust. According to recent publications, the most frequently described HASTE sequences are triggered T2-weighted TSE sequences with extended echo-train T2-weighted HASTE sequences. The thickness of the slices should not exceed 5 mm. Pulse triggering is required. Additional respiratory triggering can be done; however, this will increase the time to approximately 30 min [8]. Ventilation and perfusion maps can be generated using Fourier decomposition MRI. Ventilation maps are also produced via multi-volume acquisitions followed by measurement of signal intensities linked to inhalation. This is especially beneficial for children who suffer from severe asthma [9]. Use T1-weighted two- or three-dimensional GRE sequences instead, especially when investigating older children. Due to the short duration of the individual sequence, breath-hold acquisitions become possible. Signal loss due to susceptibility variations can be significantly decreased by using appropriate parameters (low echo time of TE < 1.0 ms) [8].
CT scoring systems exist to grade findings and monitor disease progression. A unique score for MRI does not yet exist and is still being developed. So far, the descriptive report and comparative analysis of images has been the only way of assessing the course of disease [10].
Since 1991, Bhalla scoring system has been used for HRCT scoring of CF lung disease [11], with updates to include other HRCT abnormalities [12, 13]. In the current study, the fourteen parameters of the modified Bhalla scoring system described by Judge et al. were evaluated [5]. The mean modified Helbich–Bhalla score for HRCT was 11.2 (range 3–21, SD 5.3) and that for MRI was 10.46 (range 4–20, SD 4.4), with the mean difference of 0.74 points.
In CLD as CF, many authors found a strong association between CT and MRI [10, 14, 15]. In CF, Teufel et al. found a strong correlation between CT and very-short echo time MRI (R = 0.87, p 0.01). The average Helbich–Bhalla score for CT was 12.2 [16].
In the current study, there was a strong positive correlation between CT and MRI, with (R = 0.8366, and significant P-value 0.00001) and an overall substantial agreement between the tests, Cohen's kappa was 0.689, and the overall agreement was 85%. These results were compared with a study done by Teufel et al. [16], who used the Helbich–Bhalla score, and they found a strong connection (R = 0.87). Müllera et al. reported a perfect agreement in ILD, with Cohen's kappa of 0.704 and a 91% overall agreement [17].
Although there was a strong correlation between the presence of abscesses or sacculation and the presence of honeycombing in this study, MRI overestimated detection of bronchiectasis, severity, peri-bronchial thickening, extent, presence of mucus plugs, and bronchial generation involvement by 3.6 mean point score in a few cases, this result was in agreement with a study done by Puderbach et al. [18].
Onho and coworkers found nearly perfect agreement in imaging honeycombing, traction bronchiectasis lung nodular, mass lesions, ground-glass, and consolidation opacities, as well as good agreement in visualizing emphysema, air bullae, and reticular opacities [19]. Teufel et al. discovered a strong correlation between peri-bronchial thickening and peripheral mucus plugging, as well as between bronchiectasis, mucus plugging, peri-bronchial thickening, and consolidation in their study [16].
By a mean point score of 4.1, HRCT exceeded MRI in detecting ground-glass opacifications, air bubbles, air trapping, emphysema, mosaic perfusion, presence of fibrosis, interlobular septal thickening, collapse or consolidation, and acinar nodules. This could be explained by the fact that the MRI SNR of normal lung parenchyma is insufficient for distinguishing normal tissue areas from areas with trapped air. A more straightforward approach could be to add a sequence with a very short TE to provide sufficient signal intensity from normal and pathological lung tissue in multiple breath-holds [16].
To reach the ultimate diagnosis, clinical data, along with HRCT and further laboratory tests as well as biopsy, were required. Although, in this study, MRI characterized the granulomatous disease lesions, however, it did not provide a definitive diagnosis, According to Kapur et al., MRI assisted in the detection of tuberculous lesions and suggested that it may be a good alternative instead of performing repeated CT scan for further evaluation of affected children [20].
Limitation of the study is regarding the selection of cases because CT scans were taken with advanced clinical indications in the majority of cases, so there might have been extensive severity and extent of the appearance of pulmonary changes in these patients and this might provide the better agreement of the results. However, considering a strong correlation between cross-sectional imaging and final diagnosis, we can generally assume that we have a reliable predictive value.