The advancement of MRI in differentiating Modic type I degenerative changes from early spinal infections

Abstract

MRI is the most sensitive and specific imaging method for the detection of advanced spinal infections. However, the differential diagnosis of early spinal infection and Modic Type I degenerative changes based on conventional MRI is difficult clinically, as they both may mimic each other by showing hypointensity on T ₁ weighted images and hyperintensity on T ₂ weighted spine MRI images. This review summarizes recent advancements in MRI, which may be useful in discriminating degenerative Modic Type I endplate changes from early spinal infection, and evaluates the diagnostic accuracy and limitations of MRI. We aim to provide indications for early differential diagnosis to help initiate appropriate treatment in a timely manner so that associated complications can be avoided.

Introduction

Spinal infection, which is a threatening inflammatory disease of the vertebral body and disk, may extend to the epidural or paravertebral soft tissues ¹ and constitutes 2–7% of all musculoskeletal system infections. ² Gender-specific surveys have reported that males are more likely to be affected than females (at a ratio of approximately 2:1) and that all age groups are at risk. ^3,4 The overall incidence ranges from 1:20,000 to 1:100,000, with a mortality rate of approximately 2–20%. ⁵ However, over the past few decades, the incidence of spinal infections has increased annually, and additionally, the numbers of elderly patients with chronic diseases, intravenous drug users and those undergoing spinal surgery have increased. ⁶ Early spinal infections are insidious and are characterized by a slow onset, as patients usually present with non-specific symptoms such as low back pain (85%) and fever (48%), which makes diagnosis extremely challenging. ^7–9

MRI has become the preferred imaging modality for the clinical diagnosis of advanced spinal infections including spondylitis and spondylodiscitis due to its high sensitivity (93–96%) and specificity (92–97%). ^10–12 Advanced spinal infection when there is loss of the nucleus pulposus, a reduction in disc height, destruction of the vertebral body, endplate erosion, paravertebral or spinal canal abscesses formation, ^13–17 is usually a straightforward diagnosis.

However, for early spinal infection, subchondral MRI signal intensity changes were similar in cases of Modic I degenerative changes. ^18,19 They all present on MRI as a low signal on T ₁WI endplate and subendplate bone, as a high signal on T ₂WI, and as a high signal on T2 compression lipid sequences, ⁵ which often makes the differential diagnosis difficult. If spinal infections are not diagnosed early and treated promptly, serious complications, such as spinal deformities, nerve damage and paraplegia, can occur in approximately 1% of patients. ²⁰

Therefore, the use of MRI to differentiate early spinal infection from Modic I degenerative changes is crucial to the early diagnosis and treatment of the disease. This review summarizes recent advances in the use of MRI to differentiate between Modic I changes and early spinal infections, which can help physicians make an accurate diagnosis, distinguish spinal infections from Modic I changes, and initiate appropriate treatment in a timely manner to avoid associated complications.

Identification based on conventional MRI

Conventional MRI is the imaging modality of choice for the assessment of degenerative disc changes. Compared with other imaging modalities (e.g. plain film, CT), MRI offers the advantages of multiple planes, better soft tissue contrast and visualization of nerve elements. ²¹

In cases of infection, due to inflammation-induced bone marrow edema, typical MRI of spinal infection shows a low signal between the discs and adjacent vertebral bodies on T ₁WI, with indistinguishable endplates and discs; in contrast, T ₂WI shows a high signal between the discs and adjacent vertebral bodies. ²² The intervertebral discs and adjacent vertebral bodies exhibit a high signal on T ₂WI. Contrast-enhanced MRI reveals marginal enhancement around the intraosseous abscess with adjacent bone marrow enhancement. ²³

Similarly, due to the presence of bone marrow edema, Modic Type I degenerative changes can mimic spinal infection, as the bone marrow near the vertebral endplates in patients with Modic Type I lesions show a low signal on T ₁WI and high signal at the corresponding levels of the vertebral endplates on T ₂WI. Mild or strong contrast enhancement is seen in the affected bone marrow after intravenous contrast injection. ²⁴ Therefore, it is not possible to distinguish Modic I degenerative changes from spinal infection solely based on signal changes in the vertebral endplates and adjacent bone marrow (Figure 1).

Figure 1.

Lumbar spine MRI of a 59-year-old male patient with low back pain. (a) Sagittal TSE T ₁ weighted images show low signals in the L4-5 vertebra and the corresponding intervertebral disc is involved, with indistinguishable endplates and discs. (b) T ₂WI shows a high signal between the discs and adjacent vertebral bodies. TSE, turbo spin echo.

T ₁- and T ₂ weighted images

Researchers ²⁵ found that although both infected and degenerative discs can exhibit high T2 signal intensity, a high T2 disc signal is more frequently seen in discs that are infected than in discs with Modic Type I degenerative changes. Thus, when a high T2 signal is observed in the affected disc, the clinical diagnosis may be more biased toward infection, but the diagnostic accuracy is not high. This is in agreement with the findings of Crockett et al. ²⁶

In contrast, on T ₂ weighted images, when a lack of abnormally increased signal in the intervertebral disc and no bone destruction are seen or when an empty disc ¹⁴ or Schmorl’s nodules ²⁷ are observed. Modic Type I changes are the likely diagnosis over infection. ^21,23

Contrast-enhanced MRI

Oztekin et al ²⁴ performed contrast-enhanced MRI on patients with Modic Type I changes and spinal infections and found that both discs and endplates were enhanced. Therefore, contrast-enhanced MRI is not recommended for differential diagnosis, which is supported by the findings of Yoon et al. ²⁸

Patel et al. ²⁵ performed contrast-enhanced MRI in 49 patients and found that although disc enhancement occurs in both types of patients, disc enhancement is slightly more frequent in patients with infection than in patients with Modic Type I lesions. Crockett et al ²⁶ also reported that discs in patients with Modic I vertebral endplate changes are not enhanced on gadolinium-enhanced T ₁ weighted images, whereas discs in patients with spinal infection are enhanced on gadolinium-enhanced T ₁ weighted images.

The discs of patients with both Modic Type I changes and spinal infection are enhanced on contrast-enhanced MRI (Figure 2), and although enhancement occurs more frequently in patients with infection, this is not a distinctive feature, does not help to distinguish between infection and degeneration and should not be used in the differential diagnosis. Therefore, the routine use of contrast agents may not be cost-effective when used for the initial diagnosis of spinal infection.

Figure 2.

T ₁ weighted MR Images show low signal intensity changes near the L4-5 intervertebral disc (a) in patients with Modic Type I changes and (b) spinal infection; On post-contrast T ₁ weighted images endplates disclose signal intensity increase (c), which can also be seen in spinal infection (d).

Fat suppression T ₂ weighted sequence or STIR sequence

In addition, some researchers found that using fat-suppressed T ₂ weighted sequences or short tau inversion recovery (STIR) sequences could increase the significance of bone marrow edema and that abnormalities in intramedullary high T2 signal intensity are more pronounced in fat-suppressed T ₂ weighted or STIR sequences. ^22,29 In the study by Finkenstaedt et al, ³⁰ fat-suppressed fluid-sensitive MRI sequences were able to distinguish more Modic Type I endplate changes and exhibited a greater range than standard T ₁W/T ₂W imaging. Similarly, Schwarz-Nemec et al ³¹ found that fat-suppressed, fluid-sensitive MR images were able to show more Modic I changes: the extent of vertebral edema varied significantly among Modic I changes (mean 32%), early discitis (56%) and late discitis (92%), with a sensitivity of 79.5% for the identification of Modic Type I changes when an extent of edema ≤55% was used as a threshold.

Therefore, the percentage of the area of edema within the affected vertebral body, as measured by fat-suppressed T ₂ weighted sequences or STIR sequences, can be used to differentiate between infection and Modic Type I changes, but its discriminatory ability is limited (Figure 3). In addition, some researchers ³² found that gadolinium contrast does not add any value if no bone marrow edema is observed on STIR or fat-suppressed T ₂ weighted images.

Figure 3.

Modic Type I lesions present on MRI as a high signal on T ₂WI endplate and subendplate bone of L4-L5 (a), and as a high signal on T2 compression lipid sequences (b); spinal infection also present as a high signal on T ₂WI endplate and subendplate bone L4-L5 (c), and as a high signal on T2 compression lipid sequences (d).

In summary, despite some differences in conventional MRI features, the differentiation of early spinal infection from Modic Type I degenerative changes can be very challenging, particularly when isolated vertebral involvement is present without any soft tissue component, adjacent disc involvement or abscess formation.

In highly suspicious cases where the initial findings are unclear, CT-guided biopsy must to be performed for the final diagnosis. ³³ However, it has a low diagnostic yield ranging from 30 to 75% because of antimicrobial therapy administered to patients before the biopsy. ^23,34,35 In addition, elevated inflammatory markers such as erythrocyte sedimentation rate, leukocytosis, C-reactive protein and positive blood cultures point toward spinal infections, but it may be misdiagnosed as Modic I degenerative changes in afebrile patients. ^36,37

The differential diagnosis between spinal infection and Modic Type I lesions is therefore always based on conventional MRI performance, which is not possible and requires more sensitive, alternative imaging methods.

Identification based on DWI

In recent years, the role of diffusion-weighted imaging (DWI) has been increasingly investigated in the assessment of degenerative, infected spinal and disc pathologies. DWI is a fast MRI sequence that provides an approximation of microvascular perfusion and diffusion of water molecules in the interstitium without the injection of any contrast agent. In Modic Type I changes, exhaustion of normal bone marrow elements results in an increase in extracellular volume fraction, which generates low signal intensity in DWI. Conversely, in spinal infections, a decrease in extracellular volume due to dense infiltration of inflammatory cells may contribute to an increase in DWI signal intensity. ²⁴ This makes DWI a potentially useful tool for differentiating spinal infection from Modic Type I changes, particularly in patients who cannot undergo MRI contrast enhancement due to contraindications such as renal insufficiency and allergic reactions. ^29,38

b-value

The b-value is an important parameter that affects the quality of DWI imaging. The b-value defines the diffusion moment, and for a given b-value, the signal attenuation on the DW image results from diffusion of water molecules and blood flow microcirculation perfusion. When the b-value is 0, the image is similar to a T ₂W image. As the b-value increases, the signal from the perfusion component can be eliminated, and DWI becomes more sensitive to the detection of restricted diffusion. ²⁹

Therefore, the selection of the correct b-value is key to obtaining the correct DW MR image, as it determines the perfusion of selected areas of biological tissue and the assessment of cell density. ³⁹ Several studies have shown that the use of larger b-value can overcome the "T2 transmission" effect and enhance the sensitivity of DWI. ^{12,24,40–42} The main results of several studies are summarized in Table 1.

Table 1.

Signal intensity of the endplates with spinal infection or Modic Type 1 changes on diffusion-weighted MR sequences using different b-values

Study	b-value(s/mm2)	Group	Signal intensity
			Hyperintense	Isointense	Hypointense
Daghighi et al 41	50	Spinal infection	12 (100)	0 (0)	0 (0)
		Modic Type I lesions	31 (100)	0 (0)	0 (0)
	400	Spinal infection	12 (100)	0 (0)	0 (0)
		Modic Type I lesions	27 (87.1)	4 (12.9)	0 (0)
	800	Spinal infection	11 (91.7)	1 (8.3)	0 (0)
		Modic Type I lesions	1 (3.2)	26 (83.9)	4 (12.9)
Oztekin et al 24	150	Spinal infection	46 (100)	0 (0)	0 (0)
		Modic Type I lesions	0 (0)	0 (0)	62 (100)
Balliu et al 40	500	Spinal infection	14 (100)	0 (0)	0 (0)
Fawzy et al 42	800	Spinal infection	6 (100)	0 (0)	0 (0)
Eguchi et al 12	1000	Spinal infection	9 (100)	0 (0)	0 (0)
		Modic Type I lesions	0 (0)	0 (0)	7 (100)

Values in parentheses indicate percentages.

In summary, DW-MRI may not be sufficiently discriminatory when lower b-value (50 and 400 s/mm²) are selected to identify spinal infection and Modic Type I changes; larger b-value (≥500 s/mm²) eliminate the signal from the perfusion component, and the signal becomes more sensitive to the detection of restricted diffusion. Larger b-value can thus significantly improve DWI sensitivity to discriminate between the two.

ADC values

The apparent dispersion coefficient (ADC) values primarily reflect the diffusion of extracellular free water molecules, ⁴³ as measured by depicting regions of interest (ROIs) in abnormal tissues. ⁴⁴ On the ADC map (or ROI), higher ADC values indicate increased diffusivity of water in the sampled area. Lower ADC values correspond to a lower rate of diffusivity. ⁴⁵ In contrast, a high intensity signal on DWI indicates restricted water movement.

Since intracellular water is more restricted in movement than extracellular water, tissues with high cell density and restricted diffusion of extracellular free water molecules have lower ADC values and higher DWI signal intensities. Conversely, tissues with low cell density and low intracellular organelle content or high extracellular free water content have higher ADC values and lower DWI signal intensities. ⁴⁶ For example, in Modic Type I lesions, as normal bone marrow elements are depleted, fibrovascular tissue replaces normal bone marrow between trabeculae, which leads to an increase in water content; this in turn leads to increased extracellular volume fraction, higher ADC values and lower signal intensity on DWMR images. ²⁴ In contrast, in spinal infections, dense infiltration of inflammatory cells leads to a relative decrease in extracellular volume, which results in reduced ADC values and increased signal intensity on DW MR images ⁴⁷ (Figure 4).

Figure 4.

A patient with Modic Type I lesions show slightly higher ADC values on the L5-S1 vertebrae (a), while the L4-L5 vertebrae of infected patients showed slightly lower ADC values (b). ADC, apparent dispersion coefficient.

Range of ADC values for normal, infected and Modic Type I-lesioned vertebrae

The results of several studies have shown that the mean ADC values of normal vertebral bone marrow were 0.412 ± 0.180 × 10^–3 mm ² s^–1. ^{12,42,47–52} The main results are summarized in Table 2 and Figure 5.

Table 2.

ADC values of normal vertebral bone marrow

Study	B0(T)	b-value(s/mm2)	Number	ADC (10–3 mm2/s)
Fawzy et al 42	1.5	500, 800	40	0.310 ± 0.110
Eguchi et al 12	1.5	1000	75	0.453 ± 0.121
Pui et al 47	1.5	0, 500, 1000	103	0.360 ± 0.210
Abo Dewan et al 50	1.5	1000	96	0.500 ± 0.190
Chan et al 49	1.5	200, 500, 800, 1000	49	0.230 ± 0.050
Herrmann et al 52	1.5	50, 400, 800	88	0.499 ± 0.083
Pozzi et al 48	1.5	800	33	0.408 ± 0.235
Belykh et al 51	1.5	50, 400, 800	9	0.314 ± 0.860
Averaged values			493	0.412 ± 0.180

ADC, apparent dispersion coefficient.

Figure 5.

ADC values of normal vertebral bone marrow from the studies listed in Table 1. The shaded area with the thicker vertical line is the averaged value with standard deviation over all listed studies. ADC, apparent dispersion coefficient.

For spinal infections, the mean ADC values calculated from all individual studies was 1.306 ± 0.322 × 10^–3 mm² s^–1. ^{12,40–42,47,49,50,53,54} While for Modic Type I lesions, the mean ADC values was 1.271 ± 0.452 × 10^–3 mm ² s^–1. ^12,41,51,55 The details of the ADC values found by several studies are summarized in Table 3 and in Table 4, respectively.

Table 3.

ADC values of infected vertebral bone marrow

Study	B0(T)	b-value(s/mm2)	Number	ADC (10−3mm2/s)
Daghighi et al	1.5	50, 400, 800	12	1.310 ± 0.120
Balliu et al	1.5	0, 500	14	0.963 ± 0.491
Eguchi et al	1.5	0, 1000	9	1.067 ± 0.122
Fawzy et al	1.5	500, 800	6	0.970 ± 0.220
Pui et al	1.5	0, 500, 1000	78	1.160 ± 0.400
Abo Dewan et al	1.5	1000	22	1.520 ± 0.140
Palle et al	1.5	0, 500, 1000	128	1.400 ± 0.200
TaŞKin et al	1.5	0, 500	23	1.540 ± 0.150
Chan et al	1.5	200, 500, 800, 1000	6	0.980 ± 0.210
Averaged values			298	1.306 ± 0.322

ADC, apparent dispersion coefficient.

Table 4.

ADC values of Modic Type I lesions vertebral bone marrow

Study	B0(T)	b-value(s/mm2)	Number	ADC (10−3mm2/s)
Daghighi et al	1.5	50, 400, 800	31	1.800 ± 0.120
Eguchi et al	1.5	0, 1000	7	0.624 ± 0.316
Belykh et al	1.5	50, 400, 800	9	0.498 ± 0.139
Dagestad et al	1.5	50, 400, 800	111	1.226 ± 0.352
Averaged values			158	1.271 ± 0.452

ADC, apparent dispersion coefficient.

The ADC values of both kinds of lesion is significantly higher than the ADC of normal vertebral bone marrow shown. However, the results of several studies show that the ADC values for spinal infection and for Modic Type I lesions have a large range and overlap in distribution (Figure 6).

Figure 6.

ADC values of normal, infected and Modic Type I-lesioned vertebral bone marrow from the studies listed in Table 3 and Table 4. The pink shaded area with thicker vertical lines is the averaged value with standard deviation of the infected vertebrae in Table 3. The blue shaded area with the thicker vertical line is the averaged value with standard deviation of the Modic Type I-lesioned vertebrae in Table 4. ADC, apparent dispersion coefficient.

Taken together, these results suggest that despite the restricted diffusion due to dense infiltration of inflammatory cells in spinal infections, inflammatory cells replace bone marrow fat, which leads to higher water content. Thus, this essentially enhanced diffusion associated with inflammatory edema and cellularity still has higher ADC values than those of normal vertebral bone marrow. It is important to note that ADC values may differ between different types of spinal infections ⁵⁰ (e.g. between septic and tuberculous), but the types of infection are not differentiated in some studies and are uniformly grouped and studied as “spinal infections”, which may contribute to the wide distribution of reported ADC values of infected vertebrae.

The differential value of ADC values

Combining the results of multiple studies in “Range of ADC values for normal, infected and Modic Type I-lesioned vertebrae”, we found a large range and overlap in the distribution of ADC values between spinal infection and Modic Type I lesions. This may be due to the different experimental parameters set between different studies, but there are significant differences in ADC values in individual studies, which can be used as a basis for differential diagnosis.

Eguchi et al ¹² found that the mean bone marrow ADC values were significantly higher in infected patients than in those with Modic Type I lesions (1.067 ± 0.122 × 10^–3 mm² s^–1 vs 0.624 ± 0.316 × 10^–3 mm ² s^–1). However, Daghighi et al ⁴¹ obtained a different result: the mean ADC of patients with spinal infection was significantly lower than that of patients with Modic I changes (1.31 ± 0.12 × 10^–3 mm² s^–1 vs 1.80 ± 0.12 × 10^–3 mm ² s^–1). Additionally, based on subject working curve analysis, the best cut-off ADC value of 1.52 × 10^–3 mm² s^–1 was used in this study to distinguish spinal infection from Modic I lesions, and the AUC for the cut-off value was 0.99.

The discrepancies and completely opposite results of these studies may be because ADC values are influenced by a number of factors, such as MRI technique (sequence parameters, b-values, fat suppression), ROI size and location, type of spinal infection, and definition of Modic Type I (pure Type I or mixed Type I, II); moreover, whether antimicrobial therapy was initiated prior to the DW study may also have led to the discrepancy. ⁵⁶

Therefore, ADC values for differentiating spinal infections from Modic Type I lesions have high sensitivity and specificity, but the lack of a unified parametric standard has led to wide variation in ADC values between studies. This lack of standardization has prevented the systematic use of ADC values in the differential diagnosis of early spinal infections and Modic Type I lesions.

"Claw sign" and "amorphous enhancement signal"

In addition to signal intensity and ADC values, researchers have also reported multiple specific discoveries regarding DW sequences, the “claw sign” and the “amorphous increasing signal”, which can identify spinal infections and endplate degeneration with high precision.

On DWI, the claw sign is defined as a linear, typically paired area with clear margins of high signal located at the junction between the adjacent normal bone marrow within the vertebral body and the vascularized bone marrow close to the affected disc, which is commonly seen in Modic Type I degenerative changes. This is because degenerative disc disease is a progressive process that produces a well-defined border reaction. ²⁵ In contrast, infection usually progresses rapidly and causes diffuse diffusion abnormalities and does not produce a well-defined border zone response, and thus, the signal on DW images (amorphous enhancement) is a good indicator for the diagnosis of spinal infection. ⁵⁷

In a study conducted by Patel et al. ²⁵ In all, 97 and 100% of patients were diagnosed with Modic I degenerative changes when definite claw signs were present. When suspicious claw signs were present, 84 and 100% of patients were diagnosed with Modic I degenerative changes. When the claw sign was negative or absent, 93 and 100% of patients were eventually shown to have a spinal infection. In another study, Daghighi et al ⁴¹ performed DWI on 43 patients with an unclear diagnosis on conventional MRI and found that 31 patients with Modic Type I changes had claw signs on DWI images, whereas 12 patients with spinal infection did not. In contrast, amorphous enhancement was found only on DW MR images of patients with spinal infection, and both claw and amorphous enhancement were more pronounced with lower b-value (50 s/mm²).

Thus, the “claw sign” and “amorphous enhancement signal” on diffusion-weighted images are highly predictive and accurate in identifying Modic I degenerative lesions. In patients with Type I disc interstitial signal changes, the presence of the claw sign is highly suggestive of Modic Type I degenerative changes, while the absence of the claw sign and increased amorphous signal are strongly suggestive of spinal infection (Figure 7).

Figure 7.

Sagittal diffusion-weighted MR image in a 65-year-old female with Modic Type 1 change showed a typical claw-sign (a, white box); in a 61-year-old female with spinal infection showed a amorphous increased signal (b, white box).

Identification based on water-fat MRI

As previously mentioned, several studies have demonstrated the usefulness of DWI in identifying spinal infections and Modic Type I degenerative lesions. However, the use of DWI for bone marrow assessment still has a number of limitations, such as the absence of standardized parameters, the tendency to induce artifacts at tissue boundaries, the large quantitative error in ADC values and the low sensitivity to sclerotic bone marrow. ⁵⁸

Due to the similarity of Modic Type I changes to spinal infections, the differential diagnosis requires the use of more sensitive methods to aid differentiation, particularly in cases of early infection without obvious features. On conventional MRI and DWI, some researchers have used semi-quantitative methods such as measuring the percentage of edema in an area within the involved vertebral body and using ADC values as diagnostic criteria for differentiating Modic Type I lesions from spinal infections, but the overall diagnostic accuracy of the methods is not high.

Recent advances in the development of MRI pulse sequences have introduced chemical shift-encoding-based water–fat imaging to assess vertebral bone marrow composition, a technique that enables spatially resolved assessment of bone marrow fat (BMF) fraction. ⁵⁹ It has been shown that vertebral BMF values measured using water–fat imaging are highly correlated with histological findings. ^60,61

In a retrospective study conducted by Schmeel ⁵⁸ and Fields ⁶² respectively, both of them reported that the mean BMF for spinal infections was significantly lower than that for Modic Type I degenerative changes and they were significantly lower than for normal vertebral bodies (Figure 8). The results are summarized in the Table 5.

Figure 8.

Comparison of bone marrow fat fraction (mean ± SD) in patients with normal, spinal cord infection, and MC.Tukey’s multiple comparisons tests were used.(*：p＜0.05; **：p＜0.005）. MC, Modic change; SD, standard deviation.

Table 5.

BMF fraction of vertebral lesions and normal vertebral bodies

Study	BMF fraction (%)
	Modic Type I lesions	Spinal infection	Normal vertebral bodies
Schmeel et al	35.29 ± 17.15(31)	4.28 ± 3.12(22)	51.56 ± 17.26(53)
Fields et al	23.30 ± 3.40(26)	-	45.60 ± 2.80(525)
Averaged values	29.82 ± 14.11(57)	4.28 ± 3.12(22)	46.15 ± 6.08(578)

BMF, bone marrow fat.

Values in parentheses indicate the number of vertebrae evaluated.

In addition to water–fat MRI, single-voxel proton magnetic resonance spectroscopy (MRS) has traditionally been used to measure fat content in localized regions of the vertebral body. After accounting for the effect of short T2, the MRS-based fat fraction has good equivalence with water–fat MRI. ⁵⁹ Anik ⁶³ et al performed MRS in 15 patients with vertebral tuberculous spondylitis (TBS) and Modic Type I endplate changes (MTEC) in diseased and normal vertebral bodies and found that the mean water–lipid ratio (WLR) was higher in the TBS group than in the MTEC group (7.13 vs 3.49); although not statistically significant, this difference may help in the differential diagnosis.

The above findings can be explained by the histopathological changes in the Modic I lesion: The Modic I lesion exhibits abundant vascular granulation in the endplate and subendplate areas, where fibrovascular tissue has replaced normal bone marrow, so that the mean fat fraction is significantly lower than that of the endplate without Modic changes. However, Modic I lesions contain less water and more fat than infected vertebral bodies despite increased vascularization. Therefore, the mean fat fraction of infected spines is significantly lower than that of spines with Modic Type I degenerative changes.

The use of quantitative imaging techniques (e.g. MRS, water–fat MRI) can therefore measure the vertebral BMF fraction, thus forming the basis for a quantitative and continuous assessment of endplate marrow injury, which can inform subsequent disease progression and treatment. These techniques are also new tools for differentiating Modic Type I endplate changes from spinal infections and can provide a high degree of diagnostic accuracy.

Conclusion

In summary, conventional MRI can only readily distinguish spinal infection from Modic Type I changes when a range of aggressive imaging is performed. For DWI, the “claw sign” and “amorphous enhancement signal” are accurate indicators that can differentiate these two conditions, while ADC values are limited by the lack of uniform parametric standards. In contrast, water–fat MRI has a higher diagnostic accuracy due to its more unique quantitative imaging properties. The sensitivity and specificity of different MR sequences are listed in Table 6. Nonetheless, the differentiation of early spinal infections from Modic I degenerative changes is still difficult. In recent years, the advent of Imageomics and deep learning technologies has provided new ways of differentiating these two diseases by identifying lesion sites and extracting imaging features with high accuracy. In addition, new hybrid imaging modalities, such as PET/MRI, are being developed and used to further improve the identification of spinal infections and Modic I changes, which will help doctors make early diagnoses and initiate appropriate, timely treatment to avoid associated complications.

Table 6.

The sensitivity and specificity of different MR sequences in differentiating Modic Type I degenerative changes from early spinal infections

Study	Sequences	Parameter	Sensitivity (%)	Specificity (%)	Accuracy(%)
Modic et al	Conventional MRI		96%	92%	94%
Daghighi et al	DWI	b-value: 800 s/mm2	91.70%	96.80%	95.30%
Daghighi et al		ADC (1.52 × 10－3 mm 2 s－1)	91.70%	100%	97.70%
Patel et al		Claw sign	97%	100%	98.50%
Schmeel et al	Water–fat MRI		100%	97%	98.10%

ADC, apparent dispersion coefficient; DWI, diffusion-weighted imaging.