Developed by the Applied Chest Imaging Laboratory at Brigham and Women's Hospital (Harvard Medical School), CIP is a free, open-source workstation designed exclusively for research purposes, as it is not FDA-approved or CE-marked. However, it is an interesting tool for clinicians interested in exploring the quantitative aspects of CT chest imaging, especially at institutions that cannot afford pay-per-use similar software due to financial constraints.

CIP can be downloaded as part of a pre-configured 3D Slicer package from https://chestimagingplatform.org/, which includes the older 3D Slicer version 4.10.2 which supports some obsolete or non-functional modules in newer releases. However, for access to the latest tools and advancements, particularly AI-driven segmentation, it is recommended to use the newest version of 3D Slicer and install CIP via the Extension Manager.

Applications

CIP can be used for:

• Segmentation of lung parenchyma and nodules
• Identification of airways
• Identification of bullae/emphysema, inflated lung, infiltrated lung, collapsed lung, and thoracic vessels
• Volume quantification and spatial representation of diseased and non-diseased lung regions

Therefore, CIP has potential use in:

• Chronic obstructive pulmonary disease (COPD)
• Interstitial lung disease (ILD)
• Acute lung injury
• Extraction of radiomic features from segmented tumors or nodules

Lung CT Segmenter

Segmentation can be performed using two primary approaches available through the Lung CT Segmenter module:

• Manual mode
• AI mode

Fig 1: Lung CT Segmenter, manual mode. The segmentation was achieved by placing markups on the lung.

Fig 2: Lung CT Segmenter, AI mode. The dropdown menu allows users to select from multiple AI-powered segmentation models, including lungmask R231, lungmask LTRCLobes, lungmask LTRCLobes R231, lungmask R231CovidWeb, TotalSegmentator lung basic and TotalSegmentator lung extended.

Fig 3: Lung CT Segmenter. Segmentation was achieved using the Lungmask LTRCLobes model, highlighting the segmented lung parenchyma and airways.

Lung CT Analyzer

The Lung CT Analyzer module employs thresholding techniques and a "grow from seeds" method to classify the lung into five distinct regions of interest, based on Hounsfield Unit (HU) thresholds, which can be adjusted manually to refine segmentation. The segmented regions are superimposed on 2D views for easy visualization using standard color codes. A 3D spatial reconstruction of diseased lung segments is available, providing enhanced visualization and analysis.

Fig 4: Lung CT Analyzer module interface is shown, allowing the segmentation and quantification of lung regions based on Hounsfield unit thresholds. Thresholds are adjustable for emphysema (-1050 to -950 HU), inflated lung (-950 to -750 HU), infiltrated lung (-750 to -400 HU), collapsed lung (-400 to 0 HU) and vessels (0 to 1000 HU).

Fig 5: Lung CT Analyzer. Lung analysis consists of producing five different segmentations of lungs based on Hounsfield unit range, represented in different colors and superimposed on MPR images: Bulla/emphysema (green), inflated lung (blue), infiltrated lung (orange), collapsed lung (pink) and thoracic vessels (red).

Additionally, this tool utilizes the "Segment Statistics" module to accurately calculate the volume of each lung region, providing detailed quantitative results and enabling the creation of a comprehensive PDF report.

Fig 6: Lung CT Analyzer module. Final results and quantitative table, including volumes (ml) and percentages for functional lung, emphysema, infiltrated, collapsed and affected areas (infiltrated + collapsed areas).

Interactive Lobe Segmentation 

Interactive Lobe Segmentation is a module for manual segmentation which relies on placing fiducial points on lung fissures in CT images to segment lung lobes.

Fig 7: Interactive lobe segmentation module enables the identification and separation of individual lung lobes by utilizing fiducial points (red markers) manually placed along fissures.

Parenchyma Analysis Module

Using densitometry thresholds, this module identifies and measures areas of emphysema, gas trapping, and ILD. The module generates detailed metrics, density histograms, and visualizations for selected lung regions. 

Fig 8: Parenchyma Analysis Module. The histogram shows voxel density distributions for various regions, highlighting abnormal areas and healthy lung tissue. Detailed metrics are available in the table on the left side. Densitometry is used to quantify emphysema (calculated on inspiratory scans as the percentage of voxels below a threshold, typically -950 HU (LAA%-950) or -910 HU (LAA%-910)), air trapping (measured on expiratory scans as the percentage of voxels below -856 HU) and ILD (quantified on inspiratory scans as the percentage of voxels between -600 HU and -250 HU, usually recorded as the percent high attenuation area (HAA%-600-250))

Lung lesion Analyzer

The Lung Lesion Analyzer, available only with the old version of 3D Slicer, enables precise tumor segmentation starting from a seed point within the lesion, followed by the extraction of radiomic features. 

Fig 9: A patient with adenocarcinoma was evaluated using the Lung Lesion Analyzer. The tumor was easily segmented, and radiomic features were extracted, including those from the surrounding lung parenchyma within a 15 mm radius. The segmented lesion (blue) is visualized in axial, coronal, and sagittal CT views, as well as in 3D reconstruction. This module allows users to extract radiomic features, such as first-order statistics and texture.

Parenchyma Subtype Training

The Parenchyma Subtype Training module, available only in version 4.10.2 of Slicer, facilitates the labeling of various lung structures by manually placing fiducial points. Each fiducial records information about the structure's type, subtype, region, and artefact presence, generating exportable data for analysis. This module enables the creation of training datasets for different lung diseases.

Fig 10: Parenchyma Subtype Training module in a case of centrilobular emphysema. Fiducial points (marked in pink) are manually placed on a lung CT scan to label and identify areas of centrilobular emphysema. The selected options in the module indicate the type (emphysema), subtype (centrilobular), and anatomical region (right superior lobe).

Other modules

CIP offers other modules like Trachea Stent Planning, PAA Ratio and AV ratio. However, some of these are unavailable or not functional in newer Slicer versions, requiring the use of the old release.

Fig 11: The Trachea Stent Planning module facilitates precise segmentation and measurement of the trachea and its branches for stent planning. It supports customization for Y-shaped and T-shaped stents, allowing users to place fiducial points along specific tracheal regions. The module includes fine-tuning controls for segmentation and provides detailed measurements, such as branch lengths, radii, and angles, ensuring accurate stent fitting.

Fig 12: The PAA Ratio module determines the ratio between the diameters of the pulmonary artery and the aorta, recognized as a biomarker for predicting acute exacerbations in COPD. A PAA Ratio greater than 1 serves as an indicator of exacerbation risk.

Fig 13: The AV Ratio module employs a semiautomatic approach to calculate the ratio between the airway and the vessel. After applying the predefined rules, the measurements are presented in the table. As demonstrated in the figure (illustrating a case of ILD), an A/V Ratio exceeding 1 indicates the presence of bronchiectasis.

Fig 14: The modules available on 3D Slicer version 4.10.2, which can be downloaded from https://chestimagingplatform.org/download.html.