CT Image Reconstruction: Back Projection, Filters, MPR & 3D Rendering

Three major 3D reconstruction techniques exist in CT imaging. Surface Shaded Display (SSD) visualizes organ surfaces defined by Hounsfield Unit thresholds, requiring careful threshold selection to avoid false pathology. Maximum Intensity Projection (MIP) displays highest attenuation values along multiple cuts, ideal for visualizing contrast-enhanced vessels, bones, and dense structures. Volume Rendering (VR) provides semi-transparent 3D models of volumetric structures, enabling reliable diagnosis of complex anatomical relationships. VR is particularly useful for bones, pelvic structures, and complex vascular anatomy, though oral contrast should be avoided. Opacity tables control tissue transparency based on Hounsfield Unit values. Virtual endoscopy techniques, including virtual colonoscopy and virtual bronchoscopy, use 2D and 3D volumetric reconstruction to visualize internal organ surfaces. The Navigator technique allows real-time navigation through hollow organs like the trachea and bronchi, enabling visualization of internal pathologies including obstructions and stenosis.

Virtual endoscopy is a medical visualization technique that generates fly-through videos through internal anatomical structures like the colon, bronchial tree, and sinus cavities, using CT data segmentation, centerline determination, and advanced rendering techniques; while it offers advantages such as improved patient comfort, ability to visualize inaccessible regions, and enhanced maneuverability compared to real endoscopy, its clinical utility depends critically on high-quality isotropic CT data and experienced operators, with validation studies showing comparable diagnostic performance to optical colonoscopy for polyps over 10mm when proper acquisition parameters are met.

This comprehensive section demonstrates MIP's dual clinical applications in lung imaging. First, it shows how MIP enhances lung nodule detection by displaying vessels as branching structures while maintaining nodule morphology, combined with multi-slice overlap that increases nodule visibility (demonstrated with a nodule appearing on 6 MIP slices versus only 2 standard slices). Second, it illustrates MIP's utility in classifying nodule distributions, particularly revealing 'tree-in-bud' patterns that are obscured in thin-cut reconstructions. However, the section also addresses critical limitations: MIP struggles with perihilar nodules due to overlapping vascular structures that obscure small lesions. This balanced presentation provides a complete understanding of when MIP excels and where alternative approaches may be necessary.

Maximum Intensity Projection (MIP) displays only dense structures like bone and implants. Focal trough width adjusts from 0.2mm minimum to 26mm maximum, affecting visualization range. Cross-sectional panels display multiple tooth positions with configurable display formats. Save as JPEG exports processed images. Reset volume rotation corrects orientation disruptions. These advanced techniques enable detailed implant planning and comprehensive anatomical assessment.

The Hunter system represents a high-speed multi-dimensional nonlinear scanner developed for virtual endoscopy. It enables simultaneous visualization of catalyst and extramural growth surfaces located outside the investigated cavity, such as lymph nodes or vessels. Special navigation programs automatically define the directory of virtual scanner movement in the center of the investigated cavity. The system allows specification of analyst picture details by changing the field of vision and can create relief pictures of cavities with artificial shade parts. The virtual data representation enables transformation into standard V/H/S format for use with common video systems, including telemedicine applications. This technology is particularly useful for revealing obstructive processes of upper respiratory tracts, volumetric growth of gullet, stomach, and large intestine, as well as diseases of blood vessels and bladder.

CT image artifacts are systematic discrepancies between reconstructed images and true attenuation coefficients. Categories include physics-based artifacts (beam hardening, scattering), patient-based artifacts (motion), and scanner-based artifacts (noise). Beam hardening occurs when multiple-energy X-rays interact with materials having varying attenuation coefficients, causing significant errors in reconstructed images. Metal artifacts represent the most severe beam hardening case. Various reduction methods exist: interpolation/imputation approaches, patch-based methods, mathematical characterization and subtraction, and the popular Normalized Metal Artifact Reduction (NMAR) technique that uses normalization-enhanced interpolation.

CT image artifacts are systematic discrepancies between reconstructed images and actual structures, classified by cause (physics-based, patient-related, scanner-based, sampling) and shape (streaking, wing, clipping). Beam hardening occurs because lower-energy X-rays are preferentially absorbed, leaving only high-energy photons near object centers, causing severe streaking artifacts around metallic objects. Metal artifacts result from extreme beam hardening effects. Reduction strategies include hardware approaches (physical filtration, dual-spectrum systems) and software approaches (preprocessing, post-processing inpainting). Normalized metal artifact reduction (NMAR) normalizes sinogram data, applies linear interpolation, and multiplies back by normalization vectors to reduce interface artifacts between metal and non-metal regions.

CT image artifacts are classified into three categories: physics-based artifacts (beam hardening, scatter, undersampling), patient-based artifacts (motion artifacts from respiration, cardiac cycle, or voluntary movement causing misregistration and streak artifacts; metallic artifacts from implants or jewelry causing beam hardening and streak artifacts), and scanner-based artifacts (ring artifacts from detector calibration errors in third-generation scanners; beam hardening artifacts from polypoidal beam shape in multi-detector scanners; zebra artifacts from interpolation errors in helical scanning; staircase artifacts from wide collimation or non-overlapping reconstruction intervals). Correction techniques include proper patient positioning and immobilization, sedation for pediatric patients, shortest possible scan time, removing removable metal objects, shielding non-removable implants, increasing kVp, detector recalibration, correction software (Balancing Software for ring artifacts, Beam Hardening Correction Software), overlapping reconstruction intervals, and beta blockers for cardiac imaging.

CT artifacts are systematic discrepancies between acquired images and true anatomy, categorized into physics-based (beam hardening, volume averaging, photon starvation, blooming, undersampling), patient-based (metallic, motion, incomplete projection), and scanner-based (ring, helical, cone beam, multiplanar reformations) types. Beam hardening causes cupping and streak artifacts due to preferential absorption of low-energy photons; volume averaging occurs when different densities occupy a single voxel; photon starvation creates noisy projections from insufficient photons; blooming enlarges small dense structures due to pixel saturation; metallic artifacts produce severe streaking from unmanageable densities; motion artifacts cause double contouring; ring artifacts arise from detector miscalibration; helical artifacts distort objects during spiral scanning; cone beam effects occur with multi-section scanners; and multiplanar reformations show stair-step artifacts from thick sections. Prevention strategies include proper patient positioning, breath-hold techniques, thin acquisition sections, automatic tube current modulation, metal artifact reduction software, and modern scanner technologies.

Patient-induced CT artifacts include motion artifact (caused by voluntary or involuntary movement, producing ghosting effects), beam hardening artifact (caused by X-ray beam energy changes during tissue passage), metal artifact (caused by metallic objects creating star streaking), and out of field artifact (caused by patient not being fully enclosed in the scanning field). Remedies include patient communication, short scan times, filtration, metal removal, software correction, gantry angulation, larger FOV selection, and proper patient positioning.

ALARA (as low as reasonably achievable) should be applied from procedure start. Initial settings: 15 fps for collaterals, then 7.5-6 fps for fluoroscopy. Always use lowest dose fluoro protocol. For bulky patients, change angulation rather than increasing dose. Never film balloons; use fluoro storage. Angulation is major contributor: changing from 45 to 30 degrees reduces dose rate by 25-30%.

This segment covers CT optimization principles and positioning. The primary goal of CT optimization is achieving diagnostic image quality with minimum radiation dose, following the ALARA principle. Temporal resolution is most important for imaging rapidly moving organs like the heart, reducing motion artifacts. In skull lateral view imaging, the interpupillary line (IPL) is vertical in supine position (most common) and horizontal in erect position. The X-ray beam is vertical, and the IPL is oriented perpendicular to the beam direction. Increasing reconstruction matrix size (e.g., 512 to 1024) increases spatial resolution and reduces pixel size but increases noise.

This section covers ALARA principle and MRI techniques. The instructor explains ALARA (As Low As Reasonably Achievable) principle applies to both patients and radiation workers - using minimum radiation dose while maintaining image quality. The instructor then covers MRI techniques for brain metabolites: (1) MRS (Magnetic Resonance Spectroscopy) assesses chemical composition of brain, (2) Perfusion MRI studies blood flow, (3) Diffusion MRI identifies acute stroke, (4) MRA studies blood vessels. The instructor emphasizes that MRS is specifically used for assessing brain metabolites.

ALARA (As Low As Reasonably Achievable) is the fundamental radiation protection principle—taking the best images first to avoid retakes that double radiation dose. Patient protection strategies include: only taking x-rays based on patient history and clinical examination, using proper technique, using appropriate film speed (F speed requires less than half D speed exposure), using digital imaging (lower than any traditional film), and proper collimation. Rectangular collimation reduces patient radiation by 50% compared to round collimation. Filtration (1.5mm or 2.5mm depending on kilovoltage) is required by law.

CT technology has evolved through generations: first (single detector), second (multiple detectors), third (4-64 slices), and fourth (wide coverage up to 16 cm per rotation). Ionizing radiation produces deterministic effects (cell death) and stochastic effects (cancer risk). According to BEIR reports, approximately 1 in 100 people will develop radiation-induced cancer, with higher risk in women. The ALARA principle (As Low As Reasonably Achievable) requires balancing radiation dose with image quality. Modern wide coverage CT systems have dramatically reduced radiation doses: coronary CT angiogram now delivers approximately 0.25 mSv (100-fold reduction from conventional 25 mSv). Dose management involves justification, standardization, and optimization. CT reconstruction has evolved from Filtered Back Projection to Iterative Reconstruction to Deep Learning/AI, with deep learning removing 'plastified' appearance while maintaining quality at doses 60-80% lower than conventional protocols.