Excerpt
We read with great interest the article “Management of incidental adrenal masses: an update” written by Glazer and Mayo-Smith [
1]. The article thematically covers the topics of CT, MRI and PET/CT diagnosis of incidental adrenal tumors. We also noted that the topic of MRI, specifically, chemical shift imaging, has already been commented on and discussed [
2]. In our opinion, it seems appropriate also to extend the section that deals with unenhanced CT evaluation. As stated in the article, the measurement of mean unenhanced attenuation remains fundamental, as it allows the safe identification of fat-rich adrenal adenomas. In our opinion, the possibility of performing a CT histogram analysis inside an adrenal mass could be mentioned. This method was first described in 2003 by Bae et al. and is also based on the increased content of lipids in adrenal adenomas; the principle of the method is the quantification of pixels with an attenuation lower than 0 Hounsfield units (HU) in the manually drawn region of interest [
3]. Using a cut-off value of 10% negative pixels, CT histogram analysis achieved better sensitivity and comparable high specificity compared to the mean unenhanced attenuation [
3]. The threshold of 10% negative pixels was later successfully applied in other studies [
4‐
8]. Obviously, it does not make sense to perform a CT histogram analysis in lipid-rich adenomas; however, for adenomas with a mean unenhanced attenuation higher than 10 HU, this analysis has a sensitivity of about 50% and a specificity above 98% [
5‐
7]. In other words, almost one half of lipid-poor adenomas could be safely identified by the analysis. The best results can be obtained with lesions with a mean unenhanced attenuation ranging from 11 to 25 HU. We see the main importance of the method in indeterminate adrenal masses incidentally discovered on an abdominal or thoracic CT scan, in which these initial CT images could be used for the histogram analysis and the result confirming an adrenal adenoma can save a patient from the need to undergo a CT wash-out test, which burdens the patient with both a dose of higher radiation and the administration of an iodine contrast agent. This method has proved particularly useful to distinguish lipid-poor adenomas from pheochromocytomas, adrenal carcinomas and metastases of lung cancer [
3,
6,
7]. On the other hand, this method can give false-positive results, e.g., in metastases of renal cell carcinomas [
6]. Of course, it is well known that the result of CT histogram analysis is significantly affected by noise in the CT image [
5,
6]. Therefore, CT histogram analysis cannot be recommended for CT images with a high noise level, e.g., for low-dose CT protocols. It is good to quantify the noise by measuring the standard deviation of the CT numbers inside the region of interest [
5,
6]. In clinical practice, performing CT histogram analysis is time-consuming and not all the available software packages offer such a tool. Therefore, another principle of histogram evaluation was proposed in 2016, namely the Gaussian model-based algorithm, which does not measure the number of negative pixels but utilizes only the mean attenuation and standard deviation values and the mathematical properties of a Gaussian distribution to estimate the location of the tenth percentile [
9]. A modified formula to correct CT image noise was later published to eliminate the main disadvantage of CT histogram analysis [
10]. In addition, Clark et al. made a calculator using this modified formula freely available online, which simplifies the use of this method [
10]. Texture analysis is also becoming a popular tool in the community of radiologists and could be considered a higher level of histogram analysis; papers reporting the utilization of texture analysis in adrenal masses have already been published [
11,
12]. Although CT histogram analysis is not currently part of the official guidelines, we believe that in a review article claiming to bring the latest updates in adrenal mass management and imaging, CT histogram analysis, a Gaussian model-based algorithm or texture analysis could be at least briefly mentioned, especially as a considerable number of the above-mentioned results were published in Abdominal Radiology in recent years and should not be overlooked [
5,
6,
10,
11]. We hope that our contribution could help to extend and refine the discussion of recent approaches in adrenal imaging. …
