Because the probability of malignancy is driven primarily by nodule size, accurate size estimation is crucial. The standard method for assessing nodule size is to perform uni-dimensional measurements. Such linear measurements, however, are limited by variability arising from technical factors, tumor morphology, and reader decisions [1,2]. For instance, when relying on a single unidimensional largest diameter measurement, different reviewers may select different image slices.

To remedy this shortcoming, entire tumor volume measurements have been introduced. First, with the advent of thin-section computed tomography (CT), it is now possible to acquire image data sets with spatial resolutions sufficient for accurate tumor volume measurement [3]; many recent publications have demonstrated that volumetric measurements offer better reproducibility and repeatability [4]. Second, volumetric measurement is more sensitive than unidimensional measurement in detecting even small changes. For example, in a 10 mm spherical nodule, a 1 mm increase in unidimensional diameter corresponds to a 10% increase in cross-sectional diameter and a 33% increase in volume [5]. Finally, as lung CT post-processing software becomes more widespread, tumor volumetric measurements are gaining popularity.

In this issue of the Journal of Chest Surgery, Sayan et al. [6] proposed a 3-dimensional volumetric staging system for non-small cell lung cancer as an alternative to diameter-based T category. They demonstrated that survival analysis using tumor volume yielded superior results compared to that based on diameter (p=0.04) in patients with early-stage tumors.

An interesting aspect of this study is its comparison of the actual tumor volume—determined by semi-automatic segmentation performed by 2 clinicians—with the expected volume calculated from the diameter, assuming a spherical shape and using the formula for the volume of a sphere (4/3πr³). Segmentation refers to the process by which humans (manual segmentation) or machines delineate tumor boundaries from the surrounding lung tissue [4]. Generally, the entire tumor is selected as the volume of interest; this is usually feasible but, in some cases, may be hampered by indistinct tumor margins. For example, when lung cancer is surrounded by a pathologic abnormality such as post-obstructive pneumonia, the tumor boundary is frequently obscured. From that perspective, it is both reasonable and a strength of the study that the authors included "lesion which had atelectasis accompanying the mass" as part of the exclusion criteria.

As the authors emphasize, automatic and semiautomatic segmentation methods using volumetric software have been shown to be more reproducible than manual segmentation, given that manual segmentation drawn by experts has significant drawbacks: (1) it is time-consuming, (2) it is a labor-intensive task, and (3) it has inter-reader and intra-reader variability. However, in cases of part-solid adenocarcinomas, which have a ground glass opacity (GGO) component, fully automatic segmentation can also be problematic due to the reduced contrast between the GGO component and surrounding lung parenchyma. Thus, for part-solid adenocarcinomas, semiautomatic segmentation with tumor margin editing based on a subjective decision by an experienced expert currently remains the optimal choice for accurate volumetric assessments [4]. Another point to highlight is the usage of different vendor volumetric software platforms. Studies comparing multiple volumetric software packages found considerable variation in nodule volume, indicating that the results of software packages should not be used interchangeably [7]. Another point of concern is whether nodule size is more significant than other nodule characteristics. More specifically, management guidelines on both screening-detected and incidentally discovered nodules tend to emphasize that nodule density and margin characteristics are more critical than nodule size, regardless of the method used for assessing size [7]. Ultimately, these issues remain challenging, and additional technical advances are needed.

What are the implications of these findings for clinical practice? In terms of rapid and accurate tumor segmentation, fully automatic segmentation methods based on deep learning may offer a solution. Moreover, since volumetric assessment provides a more accurate measurement of nodule size than diameter-based methods, nodule growth is likely to be assessed more precisely with volumetry. Volumetry-defined growth can be particularly useful for detecting subtle changes, especially in small but rapidly growing nodules. Further real-world studies comparing diameter-based measurements with volumetry for growth assessment are needed to solidify these findings.

In conclusion, volumetry offers practical advantages and improves imaging utility without incurring penalties in lung cancer staging.

Author contributions

All the work was done by Ho Yun Lee.

Conflict of interest

No potential conflict of interest relevant to this article was reported.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.