In the present study, using the CAT12 volume-based toolbox for SPM, and FS6 surface-based, the following evaluations were performed. First, overall differences in GMV, CT, total gray matter (TGM), and total intracranial volume (TIV) were compared in terms of ROI between both autistic and healthy groups for both methods separately. Then the regression analysis was performed for CT values and volume of TGM and TIV in both Freesurfer and CAT12 methods. Also, an agreement between the two methods was evaluated through the Bland-Altman and ICC to show that CAT12 software could be a good alternative to Freesurfer. In this study, age, VIQ, and FIQ were considered covariates in our statistical analysis due to significant differences between healthy and autism groups. In this study, age, VIQ, and FIQ were considered covariates in our statistical analysis due to significant differences between healthy and autism groups. According to studies, the effect of these variables, especially age, was not unexpected (35-37). As stated by Daniel et al.'s study, altered growth pathways in CT between the ASD group relative to TD, indicating an age-related brain structural abnormality in ASD (38). Murphy and her colleagues have demonstrated an unusual relationship between age and brain structure, such as total brain GM volume and cortical thickness (CT) (39). According to the statistical findings obtained in Tables 2 and 3, TIV and TGM were not significantly different between ASD and TDCs. Some studies reported significant differences between healthy and control TIV and TGMV groups (40-42). Still, Xiao et al. reported that there was no significant difference such as in the present study (43). Gajendra et al. calculated TIV and TGMV using the three methods of SPM, FS, and FSL. In the SPM method, there was a significant difference between the healthy and autism groups. Still, there was no significant difference in the other two methods, which could be due to overestimating by SPM compared to FSL and FS (44).
Based on the values presented in Table 4, the GM volume measured by CAT12 was significantly increased in the R inferior Middle temporal gyrus, L fusiform gyrus, and R superior frontal region of the brain in the autistic group compared to the control group. The social and problems seen in ASD may be due to deficiency in these regions, considering their role in cognition and social communication (45-47). Some studies, such as the present study, pronounced an increase (48, 49) in the volume of GM in the right inferior middle temporal gyrus region, and some, contrary to the results obtained in this study, reported a decrease (35, 50). Foster et al. shown that reduce gray matter volume in the left fusiform gyrus (18), and increases in GM volume in the left fusiform gyrus and right superior frontal in ASD have been reported previously (51, 52). Wallace et al. found an increase in the left fusiform gyrus and a reduction in the right superior frontal (52). Differences in GM volume between different age groups in autism were shown in Duerden et al. study, increasing GM volume in the left fusiform gyrus reported for children/ adolescents. Uniquely in the fusiform gyrus, was a significant likelihood of decreased gray matter volume in the adults (21). The results obtained using the freesurfer approach showed that the GM volume increases in left parsoperqularis, right parsoperqularis, right pars triangularis, left posterior cingulate, left rostral middle frontal, and right rostral middle frontal. These play important roles in social, language processing, and executive task, and structural change in any of these regions can be a reason for autism. Especially the parsoperqularis, due to its role in the human mirror nervous system (MNS), maybe a reason for social interaction deficiencies in ASD (51, 53, 54). Knaus et al. have found an increase in GM volume in parsoperqularis and triangularis (55). A significant gray matter volume reduction of both the pars opercularis and triangularis bilaterally in the subjects with ASD compared with the typical control subjects was reported by Yamasaki et al. (56). Like the present study, McAlonan et al. showed gray matter volume reduction in right triangularis (57). Duerden et al. demonstrate a GM volume increase in the posterior cingulate (21), and Skakkebaek et al. reported a decreased in it (58). Previous findings show a GM volume increment in the rostral middle frontal gyrus bilaterally (59, 60). To assess brain disorders, CT has been extensively investigated because it is one of the most sensitive biomarkers that was examined in this study by two methods based on Atlas DKT. CT differs significantly in the left caudal anterior cingulate, right cuneus, right inferior temporal, and left lateral occipital in CAT12, and left inferior parietal, left lateral occipital, and right pars orbitalis in Freesurfer. In ASD, compared to TDCs, the thickness increased in all these areas except the right cuneus. The left caudal anterior cingulate and left inferior parietal in the brain are responsible for social and cognitive behaviors (61, 62). The right cuneus and left lateral occipital are also involved in understanding visual processing (63, 64). The CT of the right inferior temporal is generally engaging in understanding language and emotions (65). A subdivision of the inferior frontal gyrus, the right pars orbitalis, is functionally associated with recognizing facial expressions of basic emotions (66). The increase in CT of left caudal anterior cingulate and right inferior temporal in the autism group compared to controls is confirmed by research (26, 67), respectively. In contradiction with the study, several prior reports have shown a CT reduction in the right inferior temporal (68), left lateral occipital (69), and right pars orbitalis (26). Zielinski et al. obtained thicker cortical during childhood in ASD in cuneus contrary, to our findings. Furthermore, they reported increased age-related cortical thinning in the inferior parietal in ASD. Also, decreased thickness in adulthood they shown in the bilateral inferior parietal by controlling for full-scale IQ (24). The inconsistency in previous studies may have been due to several factors, including data acquisition, differences in feature extraction methods and software used, mismatches of the two groups in terms of age, and Clinical, racial and climatic characteristics of the samples, as well as small sample-sizes. However, larger samples are required to achieve credible results (44, 70, 71). As the causes, risk factors, and clinical manifestations of ASD are varied, possibly related to changes in the brain's anatomical neurological abnormalities (2). In fact, ASD is associated with a variety of gene mutations, each of which affects nerve growth through a variety of pathways and methods, including gene transcription, expression and regulation, synaptic formation, and function. Cell migration can also vary the clinical manifestations of ASD symptoms (72, 73). For example, Shank/ProSAP proteins are critical to synaptic formation, function, and development. Strongly, mutations in the SHANK genes and alteration of Shank protein expression lead to abnormal synaptic development are related to learning and social cognition deficits in ASD (74). In addition, different intelligence quotients (IQ) can affect the results (75). It is better to use larger sample sizes, and inclusion criteria according to IQ, gender, age, and diagnosis in future studies. The different methods used to process and extract the feature are one of the important factors influencing the results. In this research, two approaches based on volume (CAT12) and surface-based (FS6) have been used. This article aim is to compare ROI-wise CT estimations of the CAT12 toolbox and the FS6 software for both the TDCs group and ASD. CT calculated by CAT12 showed higher values than FS6 software in all regions except the right cuneus. Regression, the Bland-Altman graphs, and ICC were used to examine the agreement and correlation between the two software. Based on the results, the regression obtained from the comparison between the two methods in terms of mean ROI of CT estimation in the TDCs group was slightly higher than the ASD group. In both TDCs and ASD groups, there was a significant relationship between the two methods according to the p-value, and considering the value of R2, a moderate correlation was obtained. In ICC analysis, the value of the F-test confirmed the study hypothesis that the relationship between the values calculated by the two software was significant, and the obtained alpha coefficient showed relatively good reliability. The Bland-Altman chart showed the average agreement between the two applications. The study by Seiger et al., Which was performed on the elderly with Alzheimer's disease, showed an excellent agreement between the CT estimates of the two methods, despite higher CT values of CAT12 than FreeSurfer (32). Also, Masouleh et al. performed thickness estimation using CAT12 toolbox and FS6 software, two large samples of healthy young and old adults, and an excellent agreement was reached between the two methods (76). Pulli et al. Reported a relatively weak agreement between the two approaches in the 5-year-old pediatric population (77). The results of the present study showed less correlation and agreement compared to the first two studies and a higher value than the study of Pulli et al. That could be because of the overestimation of CT values in the CAT12 Toolbox, the small number of samples in our study compared to these studies, and the Different age groups being considered (32, 76, 77). Also, there is diversity at different levels of analysis, and differences in analysis pipelines should be considered to estimate thickness. For example, estimates of cerebral cortex thickness, differences in algorithms used for skull stripping or brain extraction, voxel-based initial and final records, and various tissue bias correction and classification algorithms are all likely to be affected (77). Given the moderate agreement between CAT12 and FS6 approaches for the autistic group of children, it cannot be said with certainty that CAT12 Toolbox is a good alternative to FS6 software. Therefore, to achieve more credible results, in future research, it is better to consider the effects of each processing step on the results of estimating the thickness of the cerebral cortex and select a larger sample size and consider all the factors affecting the results.