The diagnostic value of magnetic resonance imaging compared to computed tomography in the evaluation of fat-containing thoracic lesions

Abstract

Intrathoracic fat-containing lesions may arise in the mediastinum, lungs, pleura, or chest wall. While CT can be helpful in the detection and diagnosis of these lesions, it can only do so if the lesions contain macroscopic fat. Furthermore, because CT cannot demonstrate microscopic or intravoxel fat, it can fail to identify and diagnose microscopic fat-containing lesions. MRI, employing spectral and chemical shift fat suppression techniques, can identify both macroscopic and microscopic fat, with resultant enhanced capability to diagnose these intrathoracic lesions non-invasively and without ionizing radiation. This paper aims to review the CT and MRI findings of fat-containing lesions of the chest and describes the fat-suppression techniques utilized in their assessment.

Introduction

Intrathoracic fat-containing lesions may arise in the mediastinum, lungs, pleura, or chest wall. CT can be helpful in the detection and diagnosis of these lesions. ¹ However, CT is solely able to identify macroscopic fat-containing lesions that contain tissue similar in attenuation to the subcutaneous fat of the chest wall. ¹ Because CT cannot demonstrate microscopic or intravoxel fat, it can fail to identify and diagnose microscopic fat-containing lesions. Such lesions may exhibit water attenuation or higher attenuation values on CT, confounding interpretation. MRI, employing spectral and chemical shift fat suppression techniques, can identify macroscopic fat and microscopic fat, respectively, with resultant enhanced capability to diagnose these intrathoracic lesions non-invasively. ² An additional advantage of MRI over CT in depicting fat is its ability to do so without ionizing radiation. Several MRI techniques have been developed over the past few decades to handle the challenges posed by cardiorespiratory motion. In addition, inflation-adjusted MRI costs have dropped considerably over the last few decades. ^3,4 MRI’s primary remaining disadvantage compared with CT is its significantly longer image acquisition time. ³

Magnetic resonance fat suppression can be achieved by two primary means and for two different purposes ¹ : spectral fat suppression or “fat saturation” to identify the presence of macroscopic fat, and ² fat suppression by chemical shift gradient-echo imaging for identification of microscopic fat. ^5,6 On MRI, macroscopic fat within a lesion appears hyperintense on T ₁- and T ₂ weighted fast spin-echo images and on in-phase T ₁ weighted images. When spectral fat suppression is utilized, the macroscopic fat signal is suppressed, an occurrence often referred to as “fat-saturation.” Spectral fat suppression can also be employed to enhance the soft tissue contrast of non-macroscopic fat-containing tissues, highlighting the T2-hyperintensity of non-fatty lesions and magnifying the enhancement of lesions on post-contrast T ₁ weighted imaging. When in- and opposed-phase chemical shift MRI is employed, areas within a lesion containing intravoxel fat and water that are imperceptible by CT, so-called “microscopic fat,” suppress on opposed-phase imaging. The term “fat-suppression” is correct terminology for both scenarios, though it is more commonly used in the latter context, with “fat-saturation” used more commonly in the context of spectral fat suppression. ^5–8

This paper aims to review the CT and MRI findings of various fat-containing lesions of the chest and describe the fat-suppression techniques used in their assessment.

Fat suppression techniques

Fat suppression is achieved via several techniques, each with specific capabilities. ^5–8 Fat suppression techniques take advantage of the difference in resonance frequency between protons in fat and water milieus. Fat–water separation methods are based on: (1) the water–fat chemical shift (resonance frequency difference); (2) the short T1 of fat; or (3) both (hybrid techniques).

Chemical-shift techniques

Two-dimensional in- and opposed-phase chemical shift MRI. There are two forms of chemical shift artifact: Type 1 and Type 2. Type 1 occurs along the axis of the frequency-encoding gradient on an MR image. ⁶ As spatial encoding by MRI depends on inferring the spatial position from the precession frequency, a water proton and a fat proton located at the same physical position will be mildly shifted within an image due to their different precession frequencies. This spatial misregistration creates a shift in the spatial location in the frequency-encoding direction and can occur not only with gradient echo imaging but also with spin echo imaging. ^6,9 This artifact appears as a dark and/or white bands at the interface between water-containing soft tissue and fat. Type 2 chemical shift artifact, also known as phase cancellation, occurs only with gradient echo sequences and is related to the phase shift between water and lipid protons. It solely occurs in the phase-encoding direction.

In- and opposed-phase MRI takes advantage of Type 2 chemical shift artifact to identify the presence of intravoxel fat (so-called microscopic fat) and water within tissue. Images are acquired using T ₁ weighted gradient-recalled echo (GRE) sequences. By selecting the appropriate time-to-echo (TE), the signal from water and fat protons in the same voxel can be made to interact constructively (in-phase TE) or destructively (opposed-phase TE). Destructive interference yields nulling of the signal within voxels containing both fat and water. By its very nature, opposed-phase imaging will not suppress or null signal within voxels containing primarily adipocytes. Therefore, this technique will not suppress the signal from adipose tissue or macroscopic fat, ⁶ though it will null signal at the interface between macroscopic fat-containing and water-containing structures to yield a chemical shift artifact known as “India ink artifact.” In the context of chest imaging, chemical shift MRI is performed during a ≤ 20 s breath-hold, to freeze respiratory motion and help ensure good imaging quality. Depending on the MRI scanner, MRI software package, and craniocaudal length of the patient’s chest, axial coverage ranging from one-third to nearly all of the chest can be achieved in a single 20 s breath-hold. The diagnostic utility of this technique lies in its sensitivity for detection of small amounts of fat in a lesion or organ, imperceptible to the eye on CT and non-chemical shift MRI, including that in adrenal adenomas, hepatic steatosis, normal thymus, thymic hyperplasia, and microscopic fat-containing pulmonary hamartomas. ^6,10–12

The Dixon technique. The Dixon method is a three-dimensional (3D) volume acquisition technique for acquisition of in- and out-of-phase imaging. In a single breath-hold acquisition, it acquires 3D in- and opposed-phase images, and then, upon post-processing, yields water-only images which suppress all macroscopic fat on the image and fat-only images which suppress all water-containing structures in the image. This technique therefore enables detection of microscopic fat via the initially acquired in- and opposed-phase T ₁ weighted images and detection of macroscopic fat via the subsequent, mathematically derived water-only images. Water-only images amplify the signal of water-containing structures while suppressing macroscopic fat. The appearance of the water-only images resembles that of spectrally fat-suppressed T ₁ weighted imaging. The Dixon method achieves both forms of fat suppression upon post-processing by summing the in-phase (water + fat) and opposed-phase (water - fat) images to produce a pure water-only image:

(water + fat) + (water - fat) = pure water-only

image and by subtracting the in- and opposed-phase images to produce a pure fat-only image:

[(water + fat) - (water - fat) = pure fat-only image ^5–8 (Figure 1).

Figure 1.

43-year-old male with pulmonary hypertension. Axial T ₁ weighted Dixon sequence with in-phase (a), opposed-phase (b), and water-only (c) images. Axial T ₂ weighted sequence with CHESS (d),STIR (e), and SPAIR (f) images showing variable degrees of fat saturation in the same patient at similar levels. In this example, the SPAIR sequence (f) yields the most selective fat-suppression due to its relative insensitivity to magnetic field inhomogeneity. CHESS, chemical shift selective; SPAIR, spectral attenuated inversion recovery; STIR, short tau inversion recovery.

Fat quantification is also possible. ¹³ The in- and opposed-phase images and fat- and water-only images acquired by the Dixon technique are obtained during a single, typically 20 s, breath-hold acquisition, allowing image co-registration and shortening of body MR imaging protocols. The Dixon technique has been used broadly for imaging of the abdomen, limbs, extremities, and spine as well. DIXON is preferred to 2D-chemical shift MR imaging for 3T MRI to ensure proper TE selection and accurate quantification of percentage signal dropout.

Disadvantages of the Dixon technique include: (1) its occasional unpredictable and inconsistent flip of water-only and fat-only post-processing within a given water-only or fat-only series of images. This water-only/fat-only post-processing swap can occur just over the lesion itself, with the remainder of the image exhibiting the appropriate, expected water or fat suppression. It is essential to be aware of this potential mishap to avert image misinterpretation/errors in tissue characterization; (2) occasionally, it too does not yield satisfactory, homogeneous macroscopic fat suppression; and (3) the image quality may be of lesser quality than standard 2D in- and opposed-phase imaging.

Spectral fat saturation (Fat-sat)/Chemical shift selective (CHESS)

Fat saturation imaging helps increase soft tissue contrast. It highlights edema and blood products on T ₁- and T ₂ weighted images and gadolinium-enhanced tissue on T ₁ weighted images by eliminating the high-intensity signal of fat; in doing so, the full spectrum of contrast gradations available at the display workstation can be applied to a narrower spectrum of contrast gradations in the acquired image data set. Macroscopic fatty tissue is extremely water-poor. Spectral fat suppression is achieved by tuning a narrow radiofrequency (RF) pulse to the resonance frequency of protons in a fatty milieu and rotating their magnetization into the transverse plane. A spoiler gradient is then applied to the transversely magnetized fat protons to dephase them. The resultant nulling of fat signal primarily leaves protons in water milieus to produce signal, highlighting non-fatty tissues, most of which contain varying amounts of water ^5–8 (Figure 1). This technique is very versatile and can be appended to any T ₁- or T ₂ weighted pulse sequence. It requires a homogeneous magnetic field to work correctly and thus may fail around metallic hardware, imaging far from isocenter, or in anatomic regions with susceptibility distortions (sinuses, head, and neck, and in some areas of the mediastinum, including the thymus and distal esophageal region). The term “CHESS” refers to “CHEmical Shift Selective” suppression in the context of spectrally selective suppression of water or fat, not in- and opposed-phase chemical shift MRI imaging. The term “fat saturation” or “fat-sat” refers specifically to spectrally selected suppression of the fat peak. The selective fat suppression version of CHESS is commonly used for T ₁ weighted contrast-enhanced MR imaging. Unlike short tau inversion recovery (STIR) imaging, another form of fat suppression to be discussed below, it is not affected by the tissue T1 value, which is shortened using intravenous contrast material. It is also employed with T ₂ weighted imaging, when increased conspicuity of the T2 signal of a lesion or soft tissue edema is desired.

Fat suppression based on the short T1 of fat

Short tau inversion recovery (STIR) imaging. STIR is another widely used sequence and is applicable at all MRI field strengths. Because fat has a shorter T1 than nearly all other tissues in the body, its signal can be selectively nulled using a magnitude-reconstructed inversion recovery sequence with short time-to-inversion (TI) values (150–180 ms at 1.5T) ^5–8 (Figure 1). This spin echo, as opposed to gradient echo, method is relatively insensitive to field inhomogeneities and can be used near metal and over large fields of view. STIR imaging has its limitations, however. The signal intensity in inversion-recovery sequences is related to the absolute value of the longitudinal magnetization (i.e. regardless of whether it has passed the null point). Therefore, tissue with a short T1 and tissue with a long T1 may produce the same signal intensity. ^5–8 The most misleading and disadvantageous aspect of STIR imaging is that fat is not the only tissue type that can be suppressed. The signal from tissue or fluid with a T1 similar to that of fat will also be suppressed, including mucus, hemorrhage, proteinaceous fluid, melanin, and gadolinium. Thus, STIR should not be used for demonstration of contrast enhancement upon intravenous gadolinium administration. Furthermore, STIR should not be used in concert with diffusion-weighted imaging (DWI) (STIR-DWI) after i.v. contrast administration. ^5–8 Lack of awareness of STIR’s disadvantages or pitfalls can lead to diagnostic error.

Hybrid techniques

SPIR and SPAIR. These techniques combine a greater-than-90° CHESS pulse with an inversion delay and spoiling to null the signal from fat. The main difference between SPIR (spectral presaturation with inversion recovery) and SPAIR (spectral attenuated inversion recovery) is that the latter uses a full 180° adiabatic inverting pulse insensitive to B0 inhomogeneity. ¹⁴ As a result, both SPIR and SPAIR are less sensitive to B0 field inhomogeneity and selectively suppress fat, in contrast to STIR that suppresses all tissues with short T1 values including fat. Also, SPIR/SPAIR sequences have higher signal-to-noise than STIR. ¹⁴ These advantages explain the increasing use of these sequences for fat suppression at MR imaging with high-field-strength magnets which are prone to greater magnetic susceptibility artifact (Figure 1).

Clinical applications in the chest

Mediastinum

Mediastinal lipomatosis and hernias involving the mediastinum

Both CT and MRI readily depict and facilitate the diagnosis of mediastinal lipomatosis and hernias involving the mediastinum, because these entities contain macroscopic fat. Mediastinal lipomatosis is a benign excessive deposition of mature adipose tissue in the mediastinum. ¹⁵ It is a benign cause of mediastinal widening on chest radiographs. Both imaging modalities will demonstrate abundant, fairly symmetrically distributed, mature, macroscopic adipose tissue in the mediastinum, without significant displacement of or mass effect on vital structures. ¹⁶

Hiatal hernias often carry perigastric fat into the mediastinum, along with adjacent stomach and occasionally other abdominal content, sometimes with a disproportionate amount of fat to bowel. Bochdalek and Morgagni hernias manifest as fatty mediastinal masses as a result of herniation of fatty intra-abdominal content through congenital defects in the posterior and anteromedial aspects of the diaphragm, respectively. ¹⁷ Retroperitoneal fat and occasionally a portion of the kidney herniate into the chest via a Bochdalek hernia. Morgagni hernias mimic paracardiac fatty masses to some extent, however the anterior abdominal fat and associated normal mesenteric vascular branches which supply it can be tracked upward through the anteromedial diaphragmatic defect into the chest along the heart, differentiating it, e.g. from a thymolipoma and a liposarcoma. Thymolipomas are not associated with a diaphragmatic defect and a liposarcoma tracking from abdomen to chest would displace normally arborizing mesenteric vessels. When the presence of a diaphragmatic defect is in question on CT, however, most commonly when overlying the isoattenuating liver or spleen in the context of traumatic diaphragmatic rupture, the higher soft tissue contrast of T ₂ weighted MRI can be employed to prove or exclude the presence of a diaphragmatic defect. On CT, the diaphragm is isoattenuating to the underlying liver and spleen. ¹ On MRI, because the skeletal muscle of the diaphragm is T2-hypointense and the liver and spleen are of intermediate T2 signal, the diaphragm is more easily discernible from these organs, as is any discontinuity of the diaphragm (Table 1).

Table 1.

Additional imaging features of fat-containing mediastinal, lung, pleural, and chest wall lesions which contribute to diagnostic specificity

Lesion	Typical imaging findings
Mediastinal lipomatosis	Abundant, symmetrically distributed, mature, macroscopic adipose tissue in the mediastinum, without significant displacement of or mass effect on vital structures.
Thymic hyperplasia	Excess thymic tissue for age, without aggressive or invasive appearance; may contain macroscopic or microscopic fat and occasionally no CT- or MR-detectable fat.
Thymolipoma	An often large, malleable, well-defined, macroscopically fatty mass in the anterior or prevascular mediastinum, which drapes along the heart.
Mature teratomas/dermoid cysts	Mass usually in the prevascular mediastinum, that may contain calcifications (latter usually T1/T2-hypointense on MRI) and fat (macro- and/or microscopic). Presence of fat-fluid level may be diagnostic.
Lipomatous hypertrophy of the interatrial septum	Non-encapsulated, well-circumscribed fatty tissue involving the interatrial septum; fat is macroscopic.
Arrhythmogenic right ventricular dysplasia	Late gadolinium enhancement on MRI and dysmotility of anterior right ventricular wall, on MRI, gated CT, or echocardiography. In MR imaging, detection of intramyocardial fat within the RV wall as signal voids employs using spectrally selective fat suppression with the black blood sequences.
Post-myocardial infarction lipomatous metaplasia	Thinned myocardial wall that presents with contraction abnormalities and macroscopic fat-containing areas within the myocardium
Extramedullary hematopoiesis	Paraspinal soft tissue mass(es), sometimes with associated rib expansion; fat may be macroscopic or microscopic
Pulmonary hamartoma	Pulmonary nodule of variable CT attenuation and MRI signal depending upon lesion content, whether fatty, cartilaginous, mixed fatty-cartilaginous, or other; fat content may be macroscopic or microscopic
Lipoid pneumonia	Chronic, fat-containing consolidative lesion with gravitational distribution; fat may be macroscopic or microscopic
Lipoma	Well-defined margins; homogeneous CT isoattenuation and MRI isointensity to subcutaneous macroscopic fat
Liposarcoma	Usually contains macroscopic fat, with or without soft tissue components; often >10 cm, with thickened septa, and nodular soft tissue, mass effect, and displacement of adjacent structures; enhancement of soft tissue component is more readily discernible by MRI than CT
Lipoblastoma	Large mediastinal lesion with septations, non-fatty and macroscopic fatty components, contrast enhancement of non-fatty component, and compartmental invasion, typically presenting in children under the age of 3 years.

Normal thymus and thymic hyperplasia

As patients age, the normal thymus is progressively replaced by fat, a process with a variable time course per individual and within the thymus itself. As fatty atrophy occurs, the thymus becomes microscopically fatty before it becomes macroscopically fatty. In patients who have a delayed time course of fatty atrophy of the thymus or who have true or lymphoid thymic hyperplasia, thymic soft tissue may be present or reappear in adulthood ^18,19 and risks misinterpretation as thymoma and lymphoma. ²⁰ Because chemical-shift MR imaging can detect microscopic fat, it can detect earlier fatty change than CT and has been used to distinguish normal thymus and thymic hyperplasia from thymic tumors (Figure 2) in most adults, but not in children and young adults whose thymuses have not yet begun to atrophy. Ascertaining the percentage signal dropout with the signal intensity index (SII) ([(lesion in-phase signal intensity - lesion out-phase signal intensity) / lesion in-phase signal intensity] * 100%) has been used to distinguish normal thymus and thymic hyperplasia from thymic tumors using a SII cut-off greater than 8.92%, with sensitivity of 100% and specificity of 100%. ¹⁸ There should be no signal dropout within thymic epithelial tumors (TETs) and lymphoma. In contrast, there is signal dropout in the normal thymus and thymic hyperplasia in most adults. Use of the SII calculation requires that the in/opposed-phase MRI acquisition be by dual echo technique. If not, the chemical shift ratio (CSR) calculation must be used, with careful ROI placement not only over the thymic tissue of interest, but also over a non-fatty chest wall muscle in the same image and avoidance of associated pitfalls. ²¹ It is important to remember that normal thymus and thymic hyperplasia in children would not be expected to suppress on opposed-phase imaging on account of insufficient fatty atrophy. Occasionally, normal and hyperplastic thymus does not suppress in adults as well. An adjunctive tool in such cases is to employ diffusion-weighted MR imaging with apparent diffusion coefficient (ADC) mapping. TETs and lymphomas will restrict water diffusion, whereas normal thymic tissue and hyperplasia will not. ^22,23

Figure 2.

37-year-old male undergoing therapy for lymphoma. Axial CT image (a) shows excess thymic soft tissue for age. Axial in-phase T ₁ weighted MR image (b) demonstrates partially fatty-intercalated T1-isointense soft tissue in the prevascular mediastinum, without compression or invasion of adjacent structures. Axial out-of-phase MR image (c) shows qualitative loss of signal intensity of this tissue, indicating the presence of microscopic fat and thereby proving the tissue to represent thymic hyperplasia.

Thymolipoma

Thymolipoma is a rare benign tumor of the thymus, histologically composed of normal thymic tissue and mature adipose tissue, with the highest incidence in young adults. Patients may be asymptomatic, present with compression symptoms, or present extremely rarely with myasthenia gravis. ²⁴ On CT, macroscopic fat and soft tissue attenuation material are demonstrated within these masses. Like CT, MRI will demonstrate the macroscopic fat within the mass (Figure 3).

Figure 3.

23-year-old male with thymolipoma. Coronal (a) and axial (b) double inversion recovery T ₁ weighted MR images and fat spin-echo T ₂ weighted axial image (c) demonstrate a large heterogeneous signal lesion in the anterior half of the chest, with predominantly high signal intensity areas representing macroscopic fat within the lesion and small scattered areas of relatively hypointense signal representing thymic tissue. On the coronal (d) and axial (e) fat-saturated T ₂ weighted images, the areas of high signal intensity completely suppress, indicating adipose tissue. The signal characteristics and draping, apparently pliable nature of this large lesion favor a thymolipoma over a liposarcoma.

Germ cell tumors

Germ cell tumors (GCTs) arise along the midline craniocaudally, from the pineal gland down to the presacral region. ²⁵ They form due to the incomplete migration of primitive germ cells during the early stage of embryonic development. ²⁵ Most GCTs arise in a gonadal tissue; however, 50–70% of extragonadal GCTs occur in the mediastinum. ²⁵ GCTs are broadly classified as teratomatous and non-teratomatous. ²⁵

Teratomas are composed of various internal constituents including fat, soft tissue, calcification, and fluid. These are easily identifiable on MRI, with the exception of calcification, which CT more reliably and definitively identifies. Intralesional macroscopic fat has high signal intensity on T ₁- and T ₂ weighted images, which is suppressed by spectral fat-suppression (Figure 4). ²⁵ When present, a fat-fluid level may be diagnostic. ²⁵ Occasionally, these tumors may contain only microscopic fat in its fluid and/or soft tissue component, which CT will fail to detect, but chemical shift MRI will demonstrate.

Figure 4.

33-year-old male with a paravertebral mediastinal teratoma. (a) Axial in-phase T ₁ weighted image demonstrates a large heterogeneous signal mass in the left paravertebral space with areas isointense to chest wall subcutaneous fat and with chemical shift India ink artifact on the (b) out-of-phase sequence delineating intralesional macroscopic fat from water-containing soft tissue. The intralesional macroscopic fat is isointense to the macroscopic fat in the chest wall on (c) the single-shot spin echo T ₂ weighted sequence and (d) the spectral fat-saturated T ₁ weighted sequence.

Intracardiac fatty lesions

Ectopic intracardiac fat is a common finding in thoracic imaging in both healthy and diseased patients. Physiologic ectopic cardiac adipose tissue may be present without clinical consequence. ²⁶ It has been observed during routine chest and cardiac CT and MR imaging and is more frequent in the right ventricle (RV) than in the left ventricle (LV), with an overall prevalence of RV ectopic intramyocardial fat of 16–43%. ²⁶ In the elderly, physiologic RV myocardial fat can be found, with linear or patchy morphology in the free wall, in the subepicardial layers of anterolateral or apical segments, and along the RV outflow tract (RVOT), in concert with preserved or increased myocardial thickness. Fat in the LV wall can indicate an old myocardial infarct.

Lipomatous hypertrophy of the interatrial septum (LHIS) is a benign non-encapsulated mass of fatty tissue infiltrating the atrial septum and was first described in 1964. ^27,28 It is typically detected in patients older than 50 years and is more common in females. It is characterized by the excessive deposition of fat in the interatrial septum, with a thickness greater than 2  cm. Although histologically benign, LHIS has been associated with adverse clinical sequelae including supraventricular arrhythmias, syncope, and sudden death. ^29,30 On imaging, it is typically dumbbell-shaped, spanning the interatrial septum but sparing the fossa ovalis. ^27,28

Arrhythmogenic right ventricular dysplasia (ARVD) is defined by fatty or fibrofatty replacement of the normal right ventricular myocardium. ³¹ The fatty replacement characteristically involves the right ventricle; however, it is occasionally present in the left ventricle. The typical patient is a young adult presenting with ventricular arrhythmia. MRI helps evaluate these patients, as it depicts right ventricular enlargement, poor right ventricular function, wall motion abnormalities, and the presence and extent of fatty or fibrofatty myocardial replacement (Figure 5). ³¹ MRI with single shot fast spin echo T ₁ weighted black blood sequences, with and without fat saturation, can facilitate the differentiation of pathological fatty infiltration from normal epicardial adipose tissue. ³¹ Late gadolinium enhancement (LGE) technique can help detect myocardial fibrous degeneration. However, it can be occasionally challenging to distinguish enhancement from intramyocardial fat in a thinned RV myocardium because both are T1-hyperintense in conventional LGE. ³² By using a multiecho Dixon fat and water separation method in LGE, distinction between fibrosis and fat is feasible. ³³ CT imaging is also used to evaluate ARVD, as it can demonstrate morphologic abnormalities such as RV enlargement, excessive trabeculations, fatty infiltration, and marked RV hypokinesis. Although cardiac CT imaging may yield high radiation doses and suboptimal volumetric assessment of the RV, it is usually used for claustrophobic patients and often used for those that receive implantable cardiac defibrillators (ICDs), though, increasingly, performance of MRI on patients with ICDs is possible if appropriate vetting of devices and other precautions are taken. ³⁴

Figure 5.

42-year-old male with ARVD and fatty infiltration of both the RV and LV. (a) Non-contrast axial CT image demonstrates areas of fatty infiltration of the myocardium in the interventricular septum. Axial cine SSFP images at diastole (b) and systole (c) demonstrate foci of fatty infiltration in the biventricular myocardium. Focal areas of wall motion abnormality (dyskinesia) are also noted (arrow). (d) Axial black blood single shot fast spin echo T ₂ weighted image shows correspondent areas of fatty infiltration (arrowheads). SAO LGE images (e, f) show patchy areas of T1 hyperintensity in the RV free wall and LV myocardium, corresponding to areas of fibrofatty infiltration. ARVD, arrhythmogenic right ventricular dysplasia; LGE, late gadolinium enhancement; LV, left ventricle; RV, right ventricle; SAO, short axis oblique; SSFP, steady-state free precession.

Post-myocardial infarction lipomatous metaplasia (PILM) is a tissue transformation process that may occur within the scar of healed myocardial infarction (MI). The prevalence of PILM at histology in the LV reached values of 68–84% in excised hearts undergoing transplantation for ischemic heart disease. ³⁵ PILM is usually found in a thinned necrotic myocardial wall that demonstrates contraction abnormalities, such as dyskinesia and hypokinesia (Figure 6). ³⁵ In cardiac MRI, PILM is generally detectable as subendocardial hyperintense stripes on EKG-gated T ₁- and T ₂ weighted black blood pulse sequences and on bright blood balanced gradient echo sequences, like cine steady-state free precession (SSFP), and on fat suppression sequences, signal will be nulled.

Figure 6.

62-year-old female with PILM. (a) Mid-ventricle short-axis CT shows thinning of the septum with lipomatous metaplasia. (b) Four-chamber late gadolinium enhancement and (c) cine SSFP images demonstrate fat within the myocardial wall of both the right and left ventricle, with chemical shift artifacts (arrows) and hyperintense T1 signal, as well as apical transmural enhancement (arrowhead). (d) Mid-ventricle short-axis late gadolinium enhancement image demonstrates transmural enhancement of the anterior wall (asterisk). Finally, two-chamber cine (e) and black-blood single shot fast spin echo T ₂ weighted imaging (f) show chemical shift artifacts (arrow) and hyperintense T2 signal within the inferior apical wall (arrowhead), respectively. PILM, post-myocardial infarction lipomatous metaplasia; SSFP, steady-state free precession.

Extramedullary hematopoiesis

Extramedullary hematopoiesis (EMH) is the production of blood elements outside of the bone marrow. EMH generally occurs in patients with deficient bone marrow hematopoiesis secondary to either peripheral red cell destruction or marrow replacement. It derives from the escape of progenitor cells from marrow which lodge in other organs. ³⁶ EMH is most often seen in the reticuloendothelial system (liver, spleen, and lymph nodes) and rarely occurs in other organs. It is often microscopic and asymptomatic, but it can sometimes yield tumor-like masses. In the thorax, it most commonly manifests on CT and MRI as paravertebral, macroscopic fat-containing masses, with occasionally expanded ribs. When no macroscopic fat is present within a paravertebral mass, the differential diagnosis broadens to include neurogenic tumors, IgG4-related disease, and lymphoproliferative lesions such as lymphoma. Chemical shift MRI, unlike CT, can help to differentiate microscopically fatty EMH tumor-like masses from the above-mentioned non-fat-containing paravertebral masses (Figure 7).

Figure 7.

83-year-old male with right paraspinal thoracic mass and biopsy-proven extramedullary hematopoiesis. MRI demonstrates a heterogeneous signal intensity lesion on all pulse sequences: (a) fat-saturated T ₁ weighted image without IV contrast; (b) fat-saturated T ₁ weighted image with IV contrast; (c) T ₁ weighted in-phase and (d) out-of-phase sequences show qualitative signal dropout or suppression on the opposed-phase image, indicating the presence of microscopic fat, and strongly favoring a diagnosis of extramedullary hematopoiesis over a lymphoproliferative process. (e) T ₂ weighted coronal image of this right paraspinal mass.

Lung

Pulmonary hamartoma

Pulmonary hamartoma is the third most frequent type of solitary pulmonary nodule and is the most common benign pulmonary neoplasm. ^2,37 Some, but not all, contain macroscopic or microscopic fat and cartilage. The radiologic diagnosis of a fat-containing pulmonary hamartoma is often made by CT detection of characteristic popcorn calcification and/or macroscopic fat within the lesion. ² Nevertheless, 50% of pulmonary hamartomas may not contain detectable fat or calcification by CT. ^2,29 MRI with thin-slice collimation (typically 3–4 mm, depending upon the size of the lesion) of a pulmonary hamartoma can not only identify macroscopic fat-containing hamartomas, which CT can do, but also identify microscopic fat-containing hamartomas. The percentage decrease in signal intensity is calculated by the SII as follows: [(SI_IP – SI_OP) / SI_IP] × 100, where SI_IP and SI_OP are the signal intensities of the nodule measured on in- and opposed-phase images. An SII greater than 17% is considered an appropriate cut-off value for lipid-rich lesions. ² Calcifications will generally manifest as low T1/T2 signal foci. ³⁰ Spectral fat suppression will allow for detection of macroscopic fat in the nodule on MRI (Figure 8). ³⁸ Cartilage within a hamartoma will demonstrate the MRI signal characteristics of water (T1-hypointense, T2-hyperintense, non-enhancing). ³⁹ On dynamic contrast-enhanced (DCE) imaging, both fat- and cartilage-containing pulmonary hamartomas typically demonstrate peripheral enhancement and enhancement of any internal clefts or septae, with little-to-no enhancement of the interstices of the lesion, because water-rich cartilage and fat do not discernibly enhance. Post-processed subtraction DCE imaging can highlight the hypoenhancement or lack of enhancement of both fat- and cartilage-containing pulmonary hamartomas, which might otherwise be less discernible, and should therefore be referenced during imaging interpretation.

Figure 8.

50-year-old male with pulmonary hamartoma. (a) Axial CT image demonstrate a heterogenous attenuation lesion containing small areas of fatty attenuation. The lesion is of heterogeneous signal on (b) fat-saturated T ₂ weighted imaging and show marked qualitative suppression of some areas between the (c) in-phase and (d) opposed-phase images, indicating the presence of microscopic fat and proving the lesion to represent a hamartoma (provided there is no history of metastatic fat-containing renal cell carcinoma, hepatocellular carcinoma, or liposarcoma).

Lipoid pneumonia

Lipoid pneumonia results from the pulmonary accumulation of endogenous or exogenous lipids. Host tissue reactions to the inhaled substances differ according to their chemical characteristics. Therefore, symptoms can vary significantly among individuals, ranging from asymptomatic to severe, life-threatening disease. ⁴⁰ Possible complications include superinfection by non-tuberculous mycobacteria, pulmonary fibrosis, respiratory insufficiency, and cor pulmonale. The disease presents with variable patterns and distribution. The finding of macroscopic fat within lung consolidation on CT (or MRI) is diagnostic of this entity. When macroscopic fat is not present within these lesions on CT, the CT findings are non-specific and can mimic many other forms of pneumonia, in addition to pulmonary neoplasms. ⁴⁰ Exogenous and endogenous lipoid pneumonia manifests as an adipose-containing mass. On CT, lipoid pneumonia may demonstrate similar attenuation values to macroscopic fat, but occasionally it contains insufficient fat to detect by CT. In addition to demonstrating fat saturation of macroscopically fatty lipoid pneumonia, MRI can demonstrate CT-occult, microscopic fat-containing lipoid pneumonia. On MRI, lesions present with high signal on T ₁ weighted images and signal loss on the spectral fat-saturated image if macroscopic fat is present and with signal loss on the opposed-phase chemical shift GRE T ₁ weighted sequence if microscopic fat is present (Figure 9).

Figure 9.

43-year-old female with lipoid pneumonia due to chronic aspiration of mineral oil. Axial CT images (a, b) demonstrate areas of bilateral lower lobe consolidation, with adjacent ground-glass and superimposed interlobular septal thickening, the latter forming the so-called “crazy paving” pattern. No definite fat is discernable within this consolidation on CT, allowing for possible overlying artifact. On MRI, microscopic fat-containing areas within the consolidation demonstrate loss of signal between the in-phase (c) and opposed-phase (d) sequences, indicating the presence of lipid within the consolidation and thereby a specific diagnosis of lipoid pneumonia.

Extrapulmonary (pleura, chest wall, mediastinum)

Lipoma, liposarcoma and lipoblastoma

Lipoma is the most frequently encountered benign soft tissue tumor and originates from adipose cells. ^41,42 Lipomas can arise in the pleura. In the chest wall, they are categorized according to their location as superficial, e.g. subcutaneous, or deep-seated, e.g. intramuscular. ^41,42 The latter can be encapsulated or infiltrative and, if infiltrative, may resemble a liposarcoma. For this reason, early radiological detection and characterization are necessary to obtain a complete wide resection and histopathological evaluation to differentiate benign from malignant lesions. CT is adequate to demonstrate the macroscopic fat in these lesions and provide a preliminary diagnosis, though arguably exposing the patient to unnecessary ionizing radiation. MRI is an excellent imaging modality to distinguish lipomas from liposarcoma, without ionizing radiation exposure. The fatty tissue in the lipomas demonstrates high signal intensity on both T ₁- and T ₂ weighted images (Figure 10). Spectral fat suppression and STIR imaging will show signal suppression of the macroscopic fat within the tumor. Lipomas can be homogeneous in signal and isointense to subcutaneous fat or heterogeneous in signal, on account of intermingled muscle fibers. ^41,42

Figure 10.

44-year-old female with a lipoma in the left serratus anterior muscle. (a) Axial T ₂ weighted image demonstrates a homogenously T2-hyperintense lesion in the left serratus anterior muscle (arrow) that is isointense to subcutaneous adipose tissue and completely suppresses on the fat-saturated T ₂ weighted sequence (b). The lesion is T1-hyperintense to muscle and isointense to subcutaneous fat (c). It suppresses on the spectrally fat-saturated T ₁ weighted sequence (d). The homogeneous, macroscopic fatty content of this lesion is compatible with a lipoma.

Primary intrathoracic liposarcomas are rare, accounting for only 2.7% of liposarcomas. ⁴³ Mediastinal liposarcomas comprise only 0.1–0.8% of all mediastinal tumors. ⁴⁴ Morphology and signal characteristics can help distinguish an infiltrative intramuscular lipoma from well-differentiated liposarcoma. For example, well-differentiated liposarcoma may show tumor tissue heterogeneity, with nodular non-adipose soft tissue, thickened, irregular, and/or nodular septa, and heterogeneous fat suppression (Figure 11). To date, the differentiation of lipomas from low-grade liposarcomas cannot be achieved exclusively by imaging CT or MRI. ⁴⁵ To our knowledge, there are no studies comparing the accuracy of CT vs MRI for this purpose. However, these imaging methods can be complimentary in the evaluation of these lesions and helpful in differential diagnostic weighting. In a series that included both CT and MRI, presence of the following imaging features favored the diagnosis of liposarcoma vs lipoma: lesion size >10 cm, percentage of fat <75%, thickened septa, presence of nodular areas, and associated non-adipose mass. ⁴⁶ In another study, MRI was 100% sensitive, 83% specific, and 84% accurate for identifying well-differentiated liposarcomas from other fatty masses. ⁴⁷ For the diagnosis of a simple lipoma, MRI was 100% specific. ⁴⁷ MRI is more sensitive than CT in detecting soft tissue enhancement within liposarcomas, on account of its higher soft tissue contrast (Figure 11).

Figure 11.

43-year-old male with pleural liposarcoma. Axial (a) and coronal (b) CT images demonstrate a large mass of heterogeneous, although primarily fatty attenuation in the posteroinferior right hemithorax, which compresses the adjacent lung. On the axial T ₂ weighted MR image (c), the lesion exhibits heterogeneous high signal intensity; its macroscopically fatty component suppresses upon spectral fat-saturation (d). On the in-phase T ₁ weighted sequence (e), the lesion demonstrates predominantly high signal intensity, with areas of suppression on the opposed-phase sequence (f) because of the presence of microscopic fat. On the axial (g) and coronal (h) fat-saturated post-contrast T ₁ weighted sequence, the lesion shows enhancement of the septae, beyond that which would be expected for enhancing muscle fibers, favoring a liposarcoma over a lipoma.

Lipoblastoma is a rare benign neoplasm of embryonal fat cells that usually presents in children. Most are diagnosed before age 3. ⁴⁸ It appears that these tumors result from the clonal expansion of mesenchymal pre-adipocytes from alterations in the pleomorphic adenoma gene 1 (PLAG1) on Chromosome 8, causing transcriptional upregulation that promotes lipoblast proliferation. ⁴⁸ Lipoblastoma most commonly occurs at sites with large amounts of immature fat in the neonate, with cases reported in the subcutaneous tissues of the extremities and the cervical, mediastinal, and peritoneal regions. ⁴⁸ Moderate complexity is frequent in lipoblastoma on MRI, with septations, non-adipose components, contrast enhancement, and compartmental invasion commonly encountered. Lipoblastoma and liposarcoma may be indistinguishable by imaging, but consideration of patient age is helpful because liposarcomas are extremely rare in children. ⁴⁹

Conclusion

MRI is a valuable tool for the evaluation of both macroscopic and microscopic fat-containing thoracic lesions without ionizing radiation exposure. The distinctive ability of MRI to detect CT-occult microscopic fat can refine the differential diagnosis and add diagnostic specificity to a broader spectrum of fat-containing lesions in the chest. Fundamental knowledge about MRI techniques, findings, and pathology can help radiologists and other healthcare providers with non-invasive diagnosis and clinical management of fat-containing intrathoracic lesions.

Highlights

• MRI of chest can identify both macroscopic fat and microscopic fat by employing spectral and chemical shift fat suppression techniques, respectively.

• Macroscopic fat-containing thoracic lesions can be diagnosed by both CT and MRI, with the latter via spectral fat suppression.

• Microscopic fat-containing thoracic lesions, including some forms of thymic hyperplasia, pulmonary hamartoma, lipoid pneumonia, and extramedullary hematopoiesis, can be more challenging to diagnose by CT, but can be specifically diagnosed by chemical shift MRI.