Volumetric measurement of cranial cavity and cerebral ventricular system with 3D Slicer software based on CT data

Abstract

Objective

This study aims to evaluate the clinical utility of using 3D Slicer software for volumetric measurement of the cranial cavity and cerebral ventricular system, particularly in hydrocephalus patients. We also provide detailed steps for performing the measurements.

Methods

Volumetric measurements were performed on 186 healthy volunteers, 117 hydrocephalus patients with intact skulls, and 72 hydrocephalus patients with incomplete skulls using 3D Slicer based on computed tomography (CT) data. CT scans were performed using a GE Discovery750 scanner and analyzed with 3D Slicer software (version 5.0.2). Cranial cavity volumes were measured using two methods: the Swiss Skull Stripper module and the Segment Editor tool. Ventricular volumes were assessed by segmenting the ventricles and periventricular structures with anatomical markers. Data were analyzed for consistency and accuracy using SPSS version 25.0, with statistical significance set at p ≤ 0.05.

Results

Intracranial volume measurements showed no significant differences between healthy controls and HANPH patients, nor between different measurement methods. In healthy controls, males had larger ventricular volumes than females, and older individuals had larger volumes, except for the fourth ventricle. The left lateral ventricle was larger than the right. No discrepancies were found between measurements taken by two neurosurgeons.

Conclusion

The volumetric measurement of cranial cavity and cerebral ventricular system with 3D Slicer software based on CT data are accurate, repeatable and consistent, providing methodological and technical support for hydrocephalus research, especially for incomplete skull patients, the third ventricle and the fourth ventricle.

Clinical trial number

Not applicable.

Introduction

Historically, intracranial volume was usually calculated by filling the cranial cavity with sand, rapeseed, or measured by cranial diameter based on computed tomography (CT) images or the head [1], while ventricle volume was obtained by cadaver anatomical studies or ventriculography. These methods, while foundational, lacked the precision and adaptability required for modern medical analysis. With advancements in imaging technology and processing software, the reconstruction and calculation of volumes for irregularly shaped tissues and organs can now be performed in three dimensions with far greater accuracy. Among these technological innovations, 3D Slicer, a free and open-source image processing software, has gained widespread use in clinical settings for the volumetric measurement of complex structures such as brain tumors [2], hypertensive cerebral hemorrhage [3], and irregular subarachnoid hemorrhages [4].

In particular, 3D Slicer has shown promise in the reconstruction of the ventricular system. Gonzalo Domínguez et al. [5] demonstrated that 3D Slicer could effectively reconstruct ventricular anatomy for educational purposes and produce measurable data on ventricle volume and periventricular structures. However, their research did not fully assess the accuracy of 3D Slicer’s volumetric measurements for the ventricular system, nor did it provide detailed step-by-step instructions for ventricle reconstruction, limiting its reproducibility and clinical application. This study aims to address these gaps by investigating the clinical value of using 3D Slicer for volumetric measurement of both the cranial cavity and the cerebral ventricular system based on CT data. Moreover, it will provide a detailed workflow for the reconstruction process to enhance its utility for both researchers and clinicians.

Materials and methods

Population

This is a cross-sectional study. From January 2020 to March 2023, adulthood people undergoing physical examination in Longyan First Affiliated Hospital of Fujian Medical University was considered as healthy volunteers. The other population was hemorrhage associated normal pressure hydrocephalus (HANPH) adulthood patients who got ventriculoperitoneal shunt in the department of Neurosurgery in Longyan First Affiliated Hospital of Fujian Medical University from September 2015 to March 2023, and was divided into with decompressive craniectomy or not. All participants provided written informed consent prior to inclusion in the study. The study was approved by the Ethics Committee of Longyan First Affiliated Hospital of Fujian Medical University.

Inclusion criteria

Healthy Volunteer Group: (1) Individuals aged 18 to 65 years. (2) No history of brain trauma, surgery, or neurological diseases. (3) Normal findings on cranial imaging and neurological examination.

HANPH Group: (1) Patients aged 18 to 65 years with a diagnosis of HANPH. (2) Underwent ventriculoperitoneal shunt surgery. (3) Complete medical records and preoperative CT data available.

Exclusion criteria

Healthy Volunteer Group: (1) Individuals with a history of brain trauma, surgery, or neurological diseases. (2) Abnormal findings on cranial imaging or neurological examination.

HANPH Group: (1) Patients with incomplete medical records or missing preoperative CT data. (2) Patients with co-existing neurological conditions unrelated to hydrocephalus, such as brain tumors or infections. (3) Patients with complications during or after ventriculoperitoneal (VP) shunt surgery, such as infection or shunt malfunction.

CT data acquisitions

All CT data were scanned from the top of the skull to the mandibular angle by GE Discovery750 CT scanner (General Electric Company, Fairfield, CT, USA) with 1.25 mm slice thickness and 1.25 mm slice interval and saved as Digital Imaging and Communications in Medicine (DICOM) format. CT DICOM of healthy volunteer was grouped as Healthy Controls; CT DICOM of HANPH without decompressive craniectomy before ventriculoperitoneal shunt was grouped as HANPH with bone; CT DICOM of HANPH without decompressive craniectomy after ventriculoperitoneal shunt 2weeks was grouped as HANPH with bone(2 W); CT DICOM of HANPH with decompressive craniectomy before ventriculoperitoneal shunt was grouped as HANPH without bone.

Preparation before measured by 3D Slicer

3D Slicer (address: https://www.slicer.org/, version 5.0.2) is downloaded and installed on a computer (the computer configuration used in this study: Win11 64-bit operating system, CPU 12th Gen Intel Core i5-12500 H, RAM 16GB). The installation address is better to use the default software installation address, not including Chinese words. CT DICOM was imported into 3D Slicer and suitable data series was selected to load (Supplementary Fig. 1A). The window level of data series was adjusted to brain window level set by 3D Slicer (Supplementary Fig. 1B). The axial level of selected data series was adjusted to Frankfurt plane (Supplementary Fig. 1C).

Volumetric measurement method of cranial cavity

Swiss Skull Stripper and Segment Editor have been proven to be useful tools for cranial cavity volumetry, offering a balance of accuracy and efficiency. While Swiss Skull Stripper is highly automated and can work well for standard cases, the Segment Editor in 3D Slicer offers higher precision at the cost of increased manual effort. Swiss Skull Stripper is more suitable for automated, large-scale studies, where speed is a critical factor. Segment Editor** offers superior control and flexibility, making it more appropriate for cases where accuracy is paramount, and the operator has sufficient expertise. In clinical and research settings, Segment Editor might have higher precision and control over segmentation are required, while Swiss Skull Stripper can be more practical in environments where automation and large-scale processing are necessary.

Volumetric measurement based on Swiss skull stripper

After preparation mentioned above, Atlas.mha and AtlasMask-label.mha were imported (download address: https://github.com/lassoan/SlicerSwissSkullStripper, Fig. 1A). The “Swiss Skull Stripper” was used step by step (Fig. 1B). After the window level of new generated data series was adjusted to brain window level (Supplementary Fig. 1B), segment_1 (renamed if you want, such as “skull volume”) was created through “Segment Editor” step by step (Fig. 1C). The holes in the segment_1 were filled through “Smoothing” (Fig. 1D). Finally, the intracranial volume was measured through “Segment Statistics” (Fig. 1E).

Fig. 1

A. Add the Atlas mask; B. Strip skull automatically by Swiss Skull Stripper; C. Put segment in new volume by threshold effects; D. Fill holes in new volume; E. Measurement of intracranial volume by Segment Statistics

Volumetric measurement based on segment editor

After preparation mentioned above, a mask was created through threshold tool whose range was from − 10 to 100, in order to cover the entire cranial cavity but exclude the skull (Supplementary Fig. 2A). Cranial cavity was painted by segment_1, while the places outside the cranial cavity, such as foramen magnum, supraorbital fissure, oval foramen, foramen, foramen lacerum, jugular foramen, where nerves and blood vessels going in and out the cranial cavity, were painted by segment_2 (Supplementary Fig. 2B). Segment_1 and segment_2 were expanded through “Grow from seeds” and corrected by “Paint” and “Erase” if not satisfied (Supplementary Fig. 2C). The holes in the segment_1 were filled through “Smoothing” (Fig. 1D) and segment_2 was deleted. Finally, the intracranial volume was measured through “Segment Statistics” (Supplementary Fig. 2D).

Volumetric measurement method of cerebral ventricular system

After preparation mentioned above, a mask was created through threshold tool whose range was from − 10 to 10 or 20, in order to cover the entire ventricle (Fig. 2A). Lateral ventricle, the 3rd ventricle, the 4th ventricle and periventricular structure, especially brain cistern were painted by four different segments respectively (Fig. 2B). These four segments were expanded through “Grow from seeds” and then segment of periventricular structure would be deleted (Fig. 2C). Monro foramen, Magendie foramen and Luschka foramen were used as anatomical markers for the boundary of each ventricle and brain cisterns. The segments of ventricle were corrected by “Paint”, “Islands” and “Erase” if not satisfied. The holes in all segments were filled through “Smoothing” (Fig. 1D) and Segment_5 was created and copied from Segment_1 through “Logical Operator” (Fig. 2D). One side of lateral ventricle in Segment_5 was cut by “Scissors” and Segment_1 was subtracted from Segment_5 through “Logical Operator” in order to separate right and left lateral ventricle (Fig. 2E). Each segment was optimized with “Wrap Solidify” to fill the holes (Fig. 2F). Finally, the ventricle volume was measured through “Segment Statistics” (Fig. 2G).

Fig. 2

A. Make a mask including cranial ventricle; B. Put four segments inside different kinds of cranial ventricle and periventricular structure separately; C. Expand segments by Grow from seeds; D. Copy segment of lateral ventricle by Logical Operator; E. Separate segment of right and left lateral ventricle; F. Fill holes in segments of ventricle by Wrap Solidify; G. Measurement of cranial ventricle volume by Segment Statistics

Statistical analysis

Volumetric measurement of cranial cavity was performed by a neurosurgeon independently using both the Swiss Skull Stripper and the Segment Editor methods. For each dataset, the measurements were taken three times using each method to ensure reliability and consistency. The average of these three measurements was used for further statistical analysis. To ensure the reliability of the measurements, especially considering the manual adjustments involved in the process using tools like “Paint,” “Erase,” and “Scissors,” a reliability study was conducted. The same neurosurgeon repeated the measurements on a random subset of 30 datasets after a period of two weeks. The intraclass correlation coefficient (ICC) was calculated to assess the reliability of the measurements. An ICC value greater than 0.75 was considered to indicate excellent reliability. For the cerebral ventricular system volumetric measurements, two neurosurgeons (both skilled in the use of 3D Slicer) performed the measurements independently. Each neurosurgeon also repeated their measurements three times, and the average of these measurements was used for statistical analysis. The reliability of the measurements between the two neurosurgeons was assessed using the ICC. Additionally, the Bland-Altman plot was used to evaluate the agreement between the measurements made by the two neurosurgeons. The Shapiro - Wilk test was used to test normality. The data with normal distribution were statistically analyzed by independent sample t-test, while data with non - normal distribution were statistically analyzed by Mann-Whitney U test or Wilcoxon rank test. Continuous variables were expressed as mean ± standard deviation. Linear regression analysis and Bland - Altman plots were used to assess the consistency of different measurement methods or neurosurgeons. Statistical analyses were performed using SPSS version 25.0 (IBM Corp.). P ≤ 0.05 was considered statistically significant.

Results

Healthy controls group included 186 cases, with average age (50.27 ± 12.03 years old) and 67.7% were female; HANPH with bone group included 117 cases, with average age (61.03 ± 11.36 years old) and 56.4% were female, the same as HANPH with bone(2 W) group; HANPH without bone group included 72 cases, with average age (52.28 ± 11.07 years old) and 50% were female.

The reliability of the intracranial volume measurements was assessed using the intraclass correlation coefficient (ICC). For the Swiss Skull Stripper method, the ICC was 0.82 (95% CI: 0.72–0.86), indicating good reliability. For the Segment Editor method, the ICC was 0.87 (95% CI: 0.80–0.92), also indicating good reliability. These results suggest that both methods provide consistent and reliable measurements of intracranial volume.

The reliability of the cerebral ventricular system measurements was assessed by comparing the measurements taken by two neurosurgeons. The ICC for the right lateral ventricle volume was 0.85 (95% CI: 0.81–0.92), for the left lateral ventricle volume was 0.88(95% CI: 0.83–0.93), for the third ventricle volume was 0.82 (95% CI: 0.76–0.87), and for the fourth ventricle volume was 0.86 (95% CI: 0.80–0.91). All ICC values were above 0.85, indicating excellent reliability between the two neurosurgeons.

For intracranial volume measurement based on Swiss Skull Stripper, independent-samples T test showed no significant difference between Healthy Controls and HANPH with bone; Mann-Whitney U test showed no significant between Healthy Controls and HANPH with bone(2 W); Wilcoxon rank test showed no significant between HANPH with bone and HANPH with bone(2 W), shown in Table 1. For two different intracranial volumetric measurement methods, there are no significant both in Healthy Controls and HANPH without bone separately (Fig. 3 and Supplementary Table 1). All kinds of ventricle volume in male are higher than those in female in Healthy Controls (Fig. 4A and Supplementary Table 2). Most kinds of ventricle volume in old people are higher than those in young people except the 4th ventricular volume (VV) in Healthy Controls (Fig. 4B and Supplementary Table 3). Left lateral ventricle volume is higher than right lateral ventricle volume in Healthy Controls no matter the gender (Fig. 4C and Supplementary Table 4). There is no difference in all kinds of ventricle volume measured by two neurosurgeons separately in Healthy Controls (Fig. 5 and Supplementary Table 5).

Table 1 Comparisons among intracranial volume based on Swiss Skull Stripper

Fig. 3

A-D: Linear regression (A, B) and Bland-Altman (C, D) of two intracranial volumetric measurement methods in Healthy Controls and HANPH without bone separately. V_SSS, Volumetric measurement based on Swiss Skull Stripper; V_SE, Volumetric measurement based on Segment Editor. The bold dashed lines represent the mean difference (bias) between the two measurement methods (VSSS and VSE). The solid lines represent the limits of agreement, which are calculated as the mean difference ± 1.96 standard deviations of the differences. The values for the mean difference and the limits of agreement are shown on the y-axis of the plots

Fig. 4

A-C: Ventricle volume compared by different kinds of factors in Healthy Controls. RLVV, right lateral ventricle volume; LLVV, left lateral ventricle volume; LVV, lateral ventricle volume; 3rd VV, the third ventricle volume; 4th VV, the fourth ventricle volume; TVV, total ventricle volume; ns, no significance; ∗, <0.05; ∗∗, <0.01; ∗∗∗, <0.001

Fig. 5

A-H: Linear regression and Bland-Altman of two neurosurgeons in Healthy Controls. RLVV, right lateral ventricle volume; LLVV, left lateral ventricle volume; 3rd VV, the third ventricle volume; 4th VV, the fourth ventricle volume; V_A Volumetric measurement by Neurosurgeon A; V_B, Volumetric measurement by Neurosurgeon B

Discussion

3D Slicer is a powerful, free, open-source medical image processing software which helps medical professionals deal with medical image data efficiently through a comprehensive set of image processing and analysis tools. A simple, convenient, repeatable and consistent volumetric measurement method of cranial cavity and cerebral ventricular system with 3D Slicer Software based on CT data was introduced in this study, especially for measurement of intracranial volume in decompressive craniectomy patients and measurement of the third ventricle volume and the fourth ventricle volume.

The intracranial volume is affected by many factors, among which genetic factors and individual development have the greatest impact. Race, one of genetic factors, is correlated strongly to region. Age is an important factor in individual development affecting the intracranial volume, but in childhood other than adulthood. Therefore, for adults in a given region, the intracranial volume is relatively stable and does not change with age. With the development and application of medical image processing software, such as 3D Slicer, ITK-SNAP, OsiriX, Amira, FreeSurfer, accurate image segmentation and volumetric measurement become popular. Table 2 shows the results of different studies on intracranial volume, for example, the average intracranial volume of Nepalese [6] was 1286.30 ± 129.88 ml; that of Koreans [7] was 1432.41 ± 162.79 ml; that of Swedish [8] was 1504 ± 169 ml; that of Chinese [9] was 1326.24 ± 95.72 ml. Considering the differences in race, sample size, age, and layer thickness of image data, the results of Chinese intracranial volume in this study (Table 1) are not significantly different from the published results above, showing the accuracy of the intracranial volumetric measurement method we used.

Table 2 Volume of cranial cavity and cerebral ventricular system for healthy people in published literatures

Some HANPH got decompressive craniectomy when first onset. However, the current studies on intracranial volumetric measurement are all from magnetic Resonance Imaging (MRI) or CT data with intact skull, leading to little knowledge about the influence of incomplete skull on automatic intracranial volumetric measurement (Swiss Skull Stripper). In our study, the consistency was demonstrated between the automatic and semi-automatic intracranial volumetric measurement (Segment Editor tool) methods in Healthy Controls through the independent sample T-test at first. Secondly, according to the paired non-parametric test, the consistency was shown between the automatic and semi-automatic intracranial volumetric measurement methods in HANPH without bone, which indicated that the intracranial volume in incomplete skull samples could be calculated accurately by automatic method based on Swiss Skull Stripper (Fig. 3).

The ventricle volume is affected by many objective factors, such as genetic factors, age, gender, diseases and external environmental factors. In addition to the above objective factors, there would be some artificial factors, such as differences in sample size, measurement methods and software used, as well as thickness of image data, leading to differences among published literatures [10,11,12,13,14,15] of normal ventricle volume (Table 2). However, there are still some rules: gender has an effect on the ventricle volume, which means the ventricle volume in male is usually higher than that in female; meanwhile the ventricle volume usually increases with age. Because of larger physique in male, the brain tissue capacity of male is larger than that of female, leading to higher ventricle volume in male. On the other hand, the ventricle would be enlarged passively due to brain atrophy in the elderly.

Comparison with the rules in published literatures above, the results obtained by 3D Slicer in this study were conformed to the general rule and not significantly different (Fig. 4A and B). In further, we found that the left ventricle was larger than the right ventricle in both men and women (Fig. 4C). Due to the fact that the volume of the left-brain tissue in most right-handed people is greater than that of the right brain tissue [16], it may be the reason why left ventricle volume was higher than the right ventricle volume. It showed that ventricle volume measurement method in this study was accurate. Because of high degree in automation and small error in repeated operation, the results of intracranial volume based on Swiss Skull Stripper were measured once and directly used for statistical analysis [17]. For ventricle volume, the function of “grow from seeds” does not always follow the ventricle boundaries during the process of automatic expansion of the threshold algorithm from a selected region to an entire area of the ventricle. Since the function of “grow from seeds” does not always follow the ventricle boundaries during the process of automatic expansion of the threshold algorithm from a selected region to an entire area of the ventricle, there may be errors in ventricle volume measurement caused by human factors when the ventricle pixels are corrected by neurosurgeons using tools such as “Paint,” “Erase,” and “Scissors.” Therefore, the average ventricle volume from the two neurosurgeons was taken as the final data for statistical analysis. If the difference between the two neurosurgeons is too large, it is necessary to recalculate the two neurosurgeon’s results and check the reason to reduce the error. However, it was repeatable and consistent between ventricle volume from two different neurosurgeons in Healthy Controls and there was a low probability of re-measurement (Fig. 5). Since the ventricle of hydrocephalus patients occupied a large proportion of the measured area, the ventricle volume measured at the ventricle boundary would not be affected seriously by the volume effect. The ventricle volume measurement should be accurate and reliable, but it may take time to manually correct. The processing time was spent mainly on manual elimination of the brain cisterns from the segmented cerebrospinal fluid.

Previous studies [6, 7, 11, 12, 18, 19] favored MRI for ventricular reconstruction due to its superior contrast between the ventricle and periventricular structures, which facilitates accurate and automated reconstruction. However, CT has traditionally struggled with this due to lower contrast and the challenges posed by heavy interstitial edema. Advances in CT technology, including thinner slices and higher resolution, have improved its ability to delineate ventricle boundaries effectively. CT offers advantages over MRI, such as shorter exam times, lower costs, and better patient tolerance, making it more suitable for frequent evaluations and long-term follow-up. Machine learning and artificial intelligence, while promising, face limitations in accurately identifying anatomical landmarks and segmenting ventricles, especially the third and fourth ventricles [14]. These technologies are still in pre-clinical stages and may not generalize well across different populations [20]. In contrast, the 3D Slicer-based volumetric measurement used in this study provides a more comprehensive and automated method for analyzing the ventricular system, including the third and fourth ventricles, which is crucial for understanding conditions like hydrocephalus.

The following aspects deserve your attention: 1. Regardless of CT or MRI data, the volumetric measurement method of cranial cavity in this study is not applicable to the data of combined head and neck scanning, such as head and neck CTA. Due to the large scanning range of such data, 3D Slicer could not automatically identify and strip the skull, and a deformed intracranial volume model would be made if forced calculated. The solution is to cut the neck data through other tools of 3D Slicer, and then use the method based on the Swiss Skull Stripper.2. Both measured by Swiss Skull Stripper method, the remaining contents with dural (epidural stripping) in CT data are called intracranial volume, while those without dural in MRI data are called brain tissue volume (subdural stripping), whose difference between the two kinds of data is the volume of the subdural space and the brain cistern. Some literatures [8] reported that adulthood intracranial volume was significantly affected by age, which should be due to the confusion of the two concepts in the measurement process. Since the growth and development of adult skull has been stable, the intracranial volume should not be greatly changed by age [14], while the brain tissue volume would be affected by brain atrophy with aging.

Despite its contributions, this study has several limitations. Firstly, the reliance on CT imaging, while practical and cost-effective, may have lower contrast resolution compared to MRI, potentially affecting the accuracy of ventricular boundary delineation. Additionally, the study’s cross-sectional design limits the ability to assess changes over time, which could be relevant for conditions like hydrocephalus that evolve. Furthermore, the study did not account for potential variability in CT scan protocols or settings between patients, which might introduce inconsistencies. Lastly, while 3D Slicer provides valuable insights, its manual adjustments and reliance on software algorithms may introduce subjectivity and variability in measurements.

Conclusion

In conclusion, the volumetric measurement of cranial cavity and cerebral ventricular system with 3D Slicer software based on CT data are accurate, repeatable and consistent, providing methodological and technical support for our subsequent hydrocephalus research. The characteristic of this method is applicable for easy and automatic measurement of intracranial volume in incomplete skull patient and ventricle volume in the third and fourth ventricles. The establishment of a widely applicable and reproducible measurement method will facilitate the comparison of similar research results and big data analysis for different research centers in the future. We hope that other researchers involved in the study of ventricle volume, especially in the field of hydrocephalus, can adopt this measurement method based on a free, open source and popular software.